feat(discovery): trigger a discovery pass on YouTube connect

A newly connected account showed no videos until the next 2h scheduled pass —
the gap that made onboarding look broken (a second user connected, saw nothing,
read as failure). The connect callback now fires an out-of-band discovery pass
for the connecting user, so videos appear promptly.

Concurrency: scheduled and connect-triggered passes share one lock (serialize),
preserving the single-fetcher invariant (ADR-018). A trigger interleaves between
the scheduler's per-user passes rather than fetching concurrently or waiting for
a whole pass. The trigger runs on the server ctx (survives the redirect) and is
non-blocking for the request goroutine.

Scope: connect-trigger only. The optional login-refresh / "Discover now" button
from #6 are intentionally not built — an unconditional login hook risks 429
storms (per the ticket's own recommendation); defer until wanted.

TDD: TestCallbackTriggersDiscovery, TestSerializeRunsOneAtATime,
TestDiscoveryTriggerEnqueueRunsUser; new BDD scenario mapped.

Refs #6

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.env.example

@@ -46,3 +46,31 @@ TAPIR_SECRETS_FILE=
 # --- run loop -------------------------------------------------------------
 # Empty/0 = single pass. Set (e.g. 15m) to poll on that cadence.
 TAPIR_POLL_INTERVAL=
+# How long to wait before re-fetching a transcript that returned HTTP 429
+# (rate_limited). Inside the window the video is skipped without hitting the
+# caption endpoint; after it expires the video is retried. 0 = always retry.
+# Go duration; default 1h.
+TAPIR_FETCH_BACKOFF=
+# Recency bound for AUTO summarization: in automatic mode only videos published
+# within this window of now are summarized; older ones are discovered + listed
+# but wait for a manual "Summarize" (so a back-catalogue doesn't self-inflict
+# 429s). An explicit request bypasses it. Go duration; default 168h (~7d).
+# 0 = no bound (summarize every unseen video).
+TAPIR_AUTO_SUMMARIZE_WINDOW=
+# Minimum interval between outbound caption fetches across the WHOLE process —
+# the shared per-egress-IP rate gate (ADR-014). Scheduler runners and the web
+# "Summarize" click-path serialise through it so they cannot collectively trip
+# 429s. Go duration; default 2s. 0 = unlimited (dev/tests).
+TAPIR_FETCH_RATE=
+
+# --- scheduled discovery (tapir serve, ADR-018) ---------------------------
+# When > 0, `serve` runs in-process discovery for ALL users on this cadence
+# (e.g. 2h): one runner pass per user per tick, run-once-on-startup then ticked.
+# Empty/0 = disabled (dev/tests never auto-fetch). SINGLE-REPLICA assumption —
+# >1 replica double-runs discovery. Go duration.
+TAPIR_DISCOVERY_INTERVAL=
+
+# --- invitations (tapir invite) -------------------------------------------
+# Public base URL used to build the invite link `tapir invite <email>` prints.
+# Default https://tapir.d-ma.be; no trailing slash needed.
+TAPIR_PUBLIC_URL=

.gitea/workflows/ci.yml

@@ -90,24 +90,4 @@ jobs:
             && echo "Smoke test passed" \
             || echo "Smoke test inconclusive: $OUTPUT"
 
-  # ── 3. Mirror to GitHub (deploy intentionally omitted until manifests exist) ─
+  # ── 3. Mirror to GitHub — skipped for now (SSH key rotation pending) ─
-  mirror:
-    name: Mirror to GitHub
-    needs: build
-    runs-on: self-hosted
-    if: github.ref == 'refs/heads/main' && github.event_name == 'push'
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-
-      - name: Push to GitHub
-        run: |
-          mkdir -p ~/.ssh
-          echo '${{ secrets.GH_DEPLOY_KEY }}' > ~/.ssh/id_rsa_gh_mirror
-          chmod 600 ~/.ssh/id_rsa_gh_mirror
-          ssh-keyscan github.com >> ~/.ssh/known_hosts 2>/dev/null
-          GIT_SSH_COMMAND="ssh -i ~/.ssh/id_rsa_gh_mirror -o IdentitiesOnly=yes" \
-            git push git@github.com:mathiasb/tapir.git HEAD:main
-          rm ~/.ssh/id_rsa_gh_mirror
-          echo "Mirrored to GitHub"

CLAUDE.md

@@ -60,8 +60,14 @@ These caused real mistakes that were caught and corrected; the corrections are l
   stores, and sinks are adapters. Adding a video provider or a sink = a new adapter implementing
   the interface, nothing in the engine changes. This is what keeps "standalone vs homelab" a
   wiring choice (ADR-003).
-- **BDD.** The `docs/use-cases/*.feature` files are the behavior spec. New behavior gets a
+- **BDD.** The `docs/use-cases/*.feature` files are the behavior spec (design records — there is
-  scenario; the use-case core is tested through fake adapters, not live YouTube/brain.
+  no godog runner). New behavior gets a scenario; the use-case core is tested through fake
+  adapters, not live YouTube/brain. A name-coverage gate (`test/acceptance/scenario_coverage_test.go`,
+  `TestScenarioCoverage`) keeps the two from drifting: every non-`@pending` scenario must be
+  mapped to an existing Go test in `scenarioCoverage`. When you add a scenario, either map it to
+  its covering test or tag it `@pending` in the `.feature` with a one-line reason. It checks the
+  *link*, not that the test exercises the scenario — that's the deliberate trade for not running
+  godog (see issue #5 / the BDD-runner decision).
 
 ## Skills (engineering discipline)
 
@@ -79,23 +85,33 @@ Skills live in the canonical library `mathias/skills` and are wired into this re
 
 ## Current build state (start here for the first task)
 
-The repo is **scaffolded and intentionally RED**:
+The repo is **green and shipping** — last tag `v0.9.0`. `task check` passes (fmt, vet, lint,
+`go test -p 1 ./...`). Go is `1.26.1` (see `go.mod`).
 
-- Clean Architecture skeleton exists: `internal/domain` (entities), `internal/ports`
+- Clean Architecture core is implemented: `internal/domain` (entities), `internal/ports`
-  (interfaces), `internal/usecase` (engine), `cmd/tapir` (entrypoint stub),
+  (interfaces), `internal/usecase.Engine.ProcessNewVideo` (resolve transcript → summarize →
-  `internal/adapters` (empty — concrete adapters go here).
+  deliver to sinks | skip on no-transcript). The acceptance tests in `test/acceptance/` are
-- `usecase.Engine.ProcessNewVideo` returns `ErrNotImplemented`.
+  green against it.
-- `test/acceptance/summarize_new_video_test.go` translates the first two Gherkin scenarios and
+- Adapters present under `internal/adapters/`: `youtube` (captions-first `VideoSource`,
-  **fails** against the stub. `task check` is therefore red on `test`.
+  timedtext/InnerTube acquisition per ADR-010), `summarizer` + `llm` (the copied AI router,
-- **First build task:** implement `ProcessNewVideo` (resolve transcript -> summarize -> deliver to
+  Primary→Fallback per ADR-004), `store` (Postgres, golang-migrate migrations 001–006),
-  sinks | skip on no-transcript) to make the acceptance tests green, following the `.feature`
+  `secrets` (file-backed `SecretStore`). The brain HTTP sink (ADR-005) is the remaining
-  files. Then add the AI-router `Summarizer` (copy `llm` per ADR-004), the YouTube `VideoSource`
+  optional sink.
-  adapter (captions-first), and the store + brain sinks.
+- Stage 1 is open (ADR-012): multi-user with **DB-enforced** isolation — Postgres RLS `FORCE`d
+  on all user-owned tables (migration 003), two-user isolation test in
+  `internal/adapters/store/rls_test.go`. Registration gate, per-user YouTube web connect, and
+  account management (disconnect / delete, ADR-013) all shipped.
+- `cmd/tapir` subcommands: `list`, `show`, `auth` (interactive host-side OAuth), `run` (batch
+  watch→summarize), `serve` (the HTMX+Templ web reader/writer under `internal/web`, a new
+  transport over the unchanged engine/ports — ADR-003). `tapir env` prints config.
+- **Build/run:** `task check` is the gate; `task build` produces the binary. Local dev uses
+  `StubAuth` (allow-all) and a `TAPIR_DB_DSN` Postgres; the deployed service uses Dex OIDC.
 
-**Unverified setup items** (see `docs/homelab-integration.md`, marked `confirm`): the Go version
+**Setup facts** (resolved — see `docs/homelab-integration.md` for the live values): LiteLLM is
-in `go.mod` (1.23 — match the koala runner; estate elsewhere uses 1.26.1), the brain-mcp URL, the
+off-cluster at `koala:30401/v1/` with `LITELLM_MASTER_KEY` from 1Password; the summarization
-exact ESO secret-ref naming, and the summarization model alias. Resolve against the live cluster
+model is config (`TAPIR_SUMMARIZER_MODEL`, default `koala/phi4-mini`), never hardcoded. The
-before depending on them, and pin answers back into `docs/homelab-integration.md`.
+brain-mcp base URL and ESO ref scheme are pinned in that doc; check it before wiring rather than
+re-deriving.
 
 ## Provenance (where this design came from)
 

DECISIONS.md

@@ -29,7 +29,9 @@ Go's concurrency model fits the watcher/worker shape).
 
 ## ADR-002 — No Supabase; reuse existing Dex / ESO / Postgres conventions
 
-**Status:** Accepted (2026-06-02)
+**Status:** Accepted (2026-06-02). The "Tapir does not write to the shared identity provider"
+implication is **partially superseded by ADR-017** (invite flow writes Dex Password CRs); the
+no-Supabase / reuse-existing-primitives decision stands.
 
 **Context.** The draft proposed self-hosted Supabase for auth + RLS + secrets (Vault). The
 homelab already runs **Dex** (OIDC), **ESO + 1Password** (secrets), and a **postgres18**
@@ -154,6 +156,22 @@ governs advancement. Reversible: if demand appears, a new ADR opens the Future C
 
 ---
 
+## ADR-009 — Trunk-Based Development
+
+**Status:** Accepted (2026-06-02)
+
+**Context.** Platform-wide convention (homelab architecture review invariant; gitea-mcp #27):
+commit directly to `main`, one logical change per commit, every commit deployable.
+
+**Decision.** Tapir follows TBD. Commit directly to `main`. No feature branches or PRs for
+solo/agent work; short-lived `agent/<desc>` branches only when parallel agents are active on
+the repo simultaneously. CI is the quality gate, not branch protection.
+
+**Consequences.** Consistent with the rest of the estate. Depends on the direct-to-main write
+path tracked in gitea-mcp #35 (item #1).
+
+---
+
 ## ADR-010 — Third-party caption acquisition via the timedtext/player baseUrl
 
 **Status:** Accepted (2026-06-02)
@@ -203,22 +221,6 @@ player/timedtext baseUrl) only. ADR-007's captions-first stance and the STT defe
 
 ---
 
-## ADR-009 — Trunk-Based Development
-
-**Status:** Accepted (2026-06-02)
-
-**Context.** Platform-wide convention (homelab architecture review invariant; gitea-mcp #27):
-commit directly to `main`, one logical change per commit, every commit deployable.
-
-**Decision.** Tapir follows TBD. Commit directly to `main`. No feature branches or PRs for
-solo/agent work; short-lived `agent/<desc>` branches only when parallel agents are active on
-the repo simultaneously. CI is the quality gate, not branch protection.
-
-**Consequences.** Consistent with the rest of the estate. Depends on the direct-to-main write
-path tracked in gitea-mcp #35 (item #1).
-
----
-
 ## ADR-011 — Web read-surface at Stage 0: Dex authn (single-user authz), action signal, public ingress + GitOps
 
 **Status:** Accepted (2026-06-02)
@@ -290,6 +292,457 @@ explicit call, with isolation as the guardrail that keeps it safe.
 
 ---
 
+## ADR-013 — Account deletion is Tapir-side only; the Dex identity is left intact
+
+**Status:** Accepted (2026-06-03). The "Tapir never holds write access to the shared identity
+provider" rationale is **partially superseded by ADR-017** (the invite flow now *creates* Dex
+Password CRs). The deletion-behaviour decision itself — delete Tapir-side state, leave the Dex
+identity intact — still stands; ADR-017 only changes the create side, not delete. See ADR-017
+for the now-asymmetric posture (Tapir can create Dex accounts but still does not delete them).
+
+**Context.** Stage 1 (ADR-012) added account deletion. A registered user is two things: a
+`users` row (plus all their data, cascade-linked) in Tapir's Postgres, and a subject identity
+in **Dex** (the homelab OIDC provider, shared across the estate — Tapir does not own it).
+"Delete my account" could mean (a) erase all Tapir-side data and secrets, or (b) that plus
+deprovision the Dex identity. The maintainer chose (a).
+
+**Decision.** Deleting a Tapir account removes **only Tapir-side state**:
+- The `users` row, cascading to all user-owned tables (`videos`, `transcripts`, `summaries`,
+  `sink_deliveries`, `video_connections`, and — via an **explicit delete**, because it has no
+  FK — `summary_actions`). The delete test asserts the cascade reaches every table and leaves
+  other users' rows untouched.
+- All of that user's secrets in the SecretStore (the per-user YouTube refresh-token refs).
+
+The **Dex identity is deliberately left intact.** Tapir does not deprovision, disable, or
+modify the shared Dex directory **on deletion**. (Note per ADR-017: Tapir *does* now create Dex
+Password CRs on invite — so the create and delete sides are deliberately asymmetric, and a
+deleted user's Dex Password CR persists. See ADR-017 consequences.)
+
+**Consequences.**
+- **Clean re-registration:** a deleted user who logs in again arrives as a Dex-authenticated
+  subject with no `users` row, so they hit the registration gate as a "new" user — no special
+  resurrection path needed. This is a feature of the choice, not an accident.
+- **Right-to-erasure is partial.** The user's *identity* still exists in Dex after deletion.
+  For Future B (trusted friends) this is acceptable: Dex is the maintainer's own directory and
+  the identity carries no Tapir content. **But if Tapir ever moves toward Future C (real
+  external/public users), this is a GDPR-shaped gap** — a true "delete my account" there must
+  also deprovision or anonymise the Dex identity, which is a new ADR and likely a Dex-admin
+  integration Tapir does not currently have. **ADR-017 widens this gap:** Tapir now *creates*
+  the Dex Password CR but does not delete it, so a deleted Tapir user leaves an orphaned Dex
+  local-password account. Tracked as a known gap in ADR-017.
+- **Blast radius:** ADR-013 originally claimed "Tapir never holds write access to the shared
+  identity provider." **ADR-017 changes this** — Tapir now holds scoped `create`+`get` on
+  `passwords.dex.coreos.com` in the `auth` namespace. The blast radius is no longer zero; it is
+  bounded by that RBAC. See ADR-017 for the security analysis.
+
+**Reversibility.** Adding Dex deprovisioning later is a superseding ADR; nothing about the
+current choice blocks it. Recorded now because "deletion is partial by design" is a deliberate
+semantic that future-Tapir (and any compliance review) must know was chosen, not overlooked.
+
+---
+
+## ADR-014 — Timedtext 429 handling: per-host backoff + honest in-flight UX, before any Whisper reconsideration
+
+**Status:** Accepted (2026-06-03). **Partially implemented as of v0.6.0** — see the
+implementation note at the end; the shared per-egress-IP rate gate (decision item 2) may not be
+fully realised. Verify against `internal/runner` + the youtube adapter before treating as done.
+**ADR-018 (in-process scheduled discovery + auto-summarize) makes decision item 2 load-bearing
+and folds its confirm/finish into that build.**
+
+**Context.** ADR-010 acquires captions from the unauthenticated `timedtext` baseUrl. Live runs
+show that endpoint **rate-limits per source IP (HTTP 429) under volume** — many videos fetched
+in one pass from one egress IP. Stage 1 (ADR-012) made this sharper in two ways: multiple users
+now drive fetches from the *same cluster egress IP*, and the v0.4.0 "Summarize" button fires an
+**immediate, synchronous-feeling** fetch on click (HTMX polls `/v/{videoId}/status`), so a 429
+now surfaces as a *user-facing stall* rather than a background batch hiccup. A throttle
+(`TAPIR_FETCH_DELAY`) exists but is a fixed inter-fetch delay, not 429-aware, and does not
+coordinate across the concurrent click-path and the `tapir run` batch path.
+
+This ADR is **not** a decision to build Whisper. ADR-007/010 keep STT deferred *pending
+measurement of the sustainable caption rate* — and that rate cannot be measured while the
+client reacts badly to the 429s it already provokes. Fix the backoff and the UX first; the
+clean data then tells you whether Whisper is warranted.
+
+**Decision.**
+
+1. **429-aware backoff at the fetch layer.** On a 429 from the timedtext/InnerTube fetch,
+   respect `Retry-After` when present; otherwise exponential backoff with jitter. This replaces
+   reliance on a fixed `TAPIR_FETCH_DELAY` alone (which stays as a floor/politeness delay).
+2. **A single per-egress-IP rate gate** shared by *both* the `tapir run` batch path and the
+   web click path, so they cannot collectively exceed the sustainable rate. Concurrency into
+   the timedtext endpoint is serialised/limited at this gate regardless of how many users or
+   goroutines are upstream. (The 429 is per *IP*, not per user — so the gate is process-/
+   cluster-egress-wide, not per-`withUser`.)
+3. **Honest in-flight UX (the product-shaping part).** The status poll distinguishes states
+   the user can understand instead of a spinner that silently stalls:
+   - *summarizing* — actively processing (the existing tapir spinner).
+   - *queued / waiting for rate limit* — fetch deferred behind the rate gate; show a calm
+     "queued, this can take a few minutes when busy" state, not a stuck spinner.
+   - *no transcript* — terminal, per ADR-010's degrade-never-error (a 429 that exhausts retries
+     resolves to `SourceNone`, same as any unavailable caption — it must not present as a hard
+     error to the user).
+   The spinner promising imminence is the wrong signal under rate-limiting; the UX must be able
+   to say "waiting" truthfully.
+4. **Measurement before Whisper.** Only once (1)-(3) are in and a real sustainable
+   per-IP rate is observed do we revisit whether caption coverage is good enough or whether the
+   deferred Whisper fallback (ADR-007) is finally warranted. That reconsideration is a future
+   ADR, gated on this data.
+
+**Consequences.**
+- Caption fetching becomes well-behaved under multi-user load instead of self-inflicting 429s;
+  the endpoint is treated as the shared, rate-limited resource it is.
+- The click-path UX stays honest: "waiting" reads as waiting, failure degrades to "no
+  transcript", never a stuck spinner or error spew.
+- A future per-IP cooldown / second egress IP / proxy becomes an option the rate gate can sit
+  in front of without UX changes.
+- **Still no Whisper** — and now there's a clean path to the *data* that decides whether it's
+  ever needed (`docs/homelab-integration.md` and a future ADR own that measurement).
+
+**Open (tracked, not in this ADR's scope):** the actual sustainable rate number; whether a
+dedicated egress IP / outbound proxy is worth it; CronJob-driven `tapir run` interaction with
+the rate gate (the batch path moves into k3s per the deferred CronJob item).
+
+**Implementation note (2026-06-03, v0.6.0 — added during reconciliation).** A v0.6.0 release
+shipped 429 handling: `domain.SourceRateLimited` (429 no longer collapsed to `SourceNone`),
+youtube adapter maps 429 → `SourceRateLimited`, **migration 007 added `videos.transcript_status`
++ `videos.rate_limited_at`** (durable per-video retry state), `TAPIR_FETCH_BACKOFF` config,
+runner records rate-limited + skips within the backoff window, and a "⏳ Retrying later" badge.
+This realises decision items 1, 3, and 4 well. **Item 2 (a single shared per-egress-IP rate
+gate) appears to be realised as per-*video* backoff state, NOT a process-wide IP gate** —
+multiple videos can still each hit the endpoint and collectively trip the per-IP 429. Treat item
+2 as **not yet confirmed done**; verify in `internal/runner`/youtube adapter and, if absent, it
+remains open. (Resolved during reconciliation: the v0.6.0 report's reference to "migration 008
+videos.rate_limited_at" was a **mislabel** — both columns shipped in migration **007**, not a
+missing 008. No migration was lost; the sequence legitimately skips 008. `rate_limited_at`
+exists and the runner's column reads are sound.) **ADR-018 makes item 2 a hard requirement:
+in-process scheduled discovery + auto-summarize drives all users' fetches through one pod egress,
+so the process-wide gate is confirmed/built as part of that slice.**
+
+---
+
+## ADR-015 — Per-user credentials: envelope-encrypted in PG18, not vault-stored
+
+**Status:** Accepted (2026-06-03)
+
+**Context.** The Stage-0/1 SecretStore (`internal/adapters/secrets/file.go`) holds per-user
+YouTube OAuth refresh tokens as a flat key-value JSON map on a PVC — explicitly a stand-in for
+"op/ESO later" (ADR-002, ADR-006). infra#86 proposed migrating it to an ESO-backed store. The
+decision spike (infra#88) found that framing subtly wrong: **ESO syncs vault→cluster at
+deploy/refresh time; it is not a runtime write API.** Per-user tokens are written *at runtime,
+per end-user* (every YouTube connect; on token rotation) — they are application state, not
+configuration. The homelab 1Password SA is also read-only, so a vault-write path would require
+a new write-capable SA, widening Tapir's blast radius to shared estate infra to store what is
+fundamentally Tapir's own row-data. Reading the actual SecretStore confirmed the shape: a
+3-method port (`Get`/`Put`/`Delete`) over opaque refs, written interactively per user.
+
+**Decision.** Per-user credentials are stored **envelope-encrypted in PG18**, not in the vault:
+
+1. Tokens are encrypted with a **single app-level envelope key** and stored as ciphertext in
+   PG18, under the Row-Level Security already enforced and tested (ADR-012). Reads/writes go
+   through the existing `withUser` RLS-scoped seam.
+2. The **envelope key** is the only secret in 1Password — fetched via the **existing read-only
+   SA** (confirmed working). No new write-capable SA; no per-user vault items.
+3. The `ports.SecretStore` port is unchanged (`Get`/`Put`/`Delete`). The implementation swaps
+   `FileStore` (PVC JSON) for a `PGStore` (encrypted rows). Every consumer — connect,
+   disconnect, delete-account — is untouched (the port abstraction holds, ADR-003 spirit).
+4. **Infra/operator credentials** (Dex client secret, MCP-auth tokens, service tokens) stay an
+   **ESO/1Password** concern. This ADR governs *per-user runtime* credentials only. The two
+   classes use two mechanisms deliberately — because they are two different things (runtime
+   app-state vs deploy-time config), not as a compromise. The "one mechanism" question
+   (maintainer's initial preference) was answered in #88 by correctly *classifying* the
+   secrets rather than unifying their storage.
+
+**Consequences.**
+- Runtime credential writes are normal RLS'd DB writes — no ESO sync latency, no indirection,
+  no write-SA blast radius. The interactive connect→store→use flow works without a vault
+  round-trip.
+- Keeps PG18 and keeps ADR-002 intact (Supabase was considered and rejected again in #88 —
+  adding a datastore to hold a few encrypted strings PG18 already holds).
+- Adds an encrypt/decrypt seam and an **envelope-key rotation** responsibility (re-encrypt the
+  per-user rows under a new key). infra#89 (build) must implement and test rotation, not assume
+  it — this is the real engineering cost of the choice.
+- The vault's involvement shrinks to one static key via the SA already trusted for reads.
+- **Supersedes** the "PVC stand-in for op/ESO" intent recorded in `secrets/file.go` and
+  `docs/homelab-integration.md` for the *per-user* secret path (the ESO/1Password reference in
+  ADR-006 stands for the *infra-cred* path).
+
+**Reversibility / falsification (from infra#88).** Revisit if: per-user tokens need
+high-frequency rotation writes (weak — PG18 handles it); an estate compliance policy requires
+all credentials in 1P for a single audit surface (maintainer-knowable, not currently believed
+to hold — would favour the vault-write path on policy grounds); or envelope-key rotation proves
+operationally worse than per-secret vault rotation (the real cost #89 must prove). If none hold,
+this stands. Full reasoning + rejected candidates (write-capable SA; Supabase): the infra#88
+decision doc (`infra/docs/superpowers/handoffs/`). Build + reboot-validation: infra#89.
+
+---
+
+## ADR-016 — Stage 0 gate revised: "useful to me OR a friend", behavioural not feedback
+
+**Status:** Accepted (2026-06-03). Revises the Stage 0 definition in VISION.md (supersedes the
+original "useful to me, specifically" gate and folds in the old Stage 1 "a trusted user returns"
+test). **The clock-start is further revised by ADR-018** (starts when scheduled discovery ships,
+since unprompted use was not possible before then).
+
+**Context.** The original Stage 0 gate was "the maintainer reads summaries weekly for four weeks
+and acts on one." The maintainer chose to change it to include friendly users, reasoning that
+early signal from friendly users is valuable. Two sub-decisions shaped the final form:
+- *Me OR a friend* (not AND): either the maintainer or an onboarded friend showing use clears it.
+- *Behavioural, not feedback*: the test is **return usage**, not stated approval.
+
+**Decision.** Stage 0 passes when, over a 3–4 week window, **either the maintainer or at least
+one onboarded friend returns to Tapir unprompted and reads/acts on summaries in ≥2 separate
+weeks.** Friend feedback is gathered and valued but is **not** the gate.
+
+**Why behavioural, not feedback (the load-bearing part).** Asked-for feedback from friendly
+users is the least reliable signal in product development — politeness bias means a friend you
+onboarded will tend to say encouraging things regardless of real value. The thing actually worth
+knowing is whether they *come back on their own*. So the gate measures returns, not nice words.
+This deliberately resists the most common way a principled gate dies: being declared "passed" on
+the strength of a polite reaction.
+
+**Honest note on what this change does.** This is a *guardrail edit made while the original gate
+was unmet* (Stage 0 had barely started; build had run well ahead of use-evidence). That is
+precisely the pattern that warrants scrutiny — redrawing a gate around work already done. It was
+examined on that basis and proceeds because: (a) the new gate is **not softer in kind** — it
+stays behavioural and sustained, merely broadening *who* can supply the signal; (b) friendly-user
+signal is genuinely valuable; (c) the politeness-bias guard keeps it from collapsing into
+"someone said it's nice." It is *not* a licence to treat the already-shipped Stage-1 machinery as
+evidence the gate passed — use-evidence remains open.
+
+**Consequences.**
+- VISION.md Stage 0 rewritten; old Stage 1 ("a trusted user returns") folded in (it was
+  near-identical to the new test); hardening renumbered to Stage 1.
+- New drift signal added: declaring the gate passed on polite feedback rather than return-usage.
+- The 2026-07-01 check-in now asks "is anyone (me or a friend) coming back unprompted?", not
+  "am I using it weekly?". (Check-in date itself revised by ADR-018.)
+
+**Reversibility.** A superseding ADR could tighten it back to maintainer-only or raise it to
+require multiple returning users. Recorded with the full rationale (including the self-scrutiny
+about editing a gate while it's unmet) so the reasoning survives, not just the new wording.
+
+---
+
+## ADR-017 — Invite flow: Tapir creates Dex local-password accounts (write access to the shared identity provider)
+
+**Status:** ~~Accepted~~ **SUPERSEDED by [ADR-019](#adr-019--authentik-owns-invites-tapir-stops-provisioning-accounts) (2026-06-07).** The Dex local-password
+invite provisioning was removed when the homelab IdP migrated Dex→Authentik
+(infra ADR-0001); Authentik now owns invites. Original record below.
+
+Accepted (2026-06-03), **recorded retroactively during reconciliation, then
+deliberately ratified KEEP (2026-06-03).** This capability **shipped in v0.7.0 without an ADR** —
+code, RBAC, and a deployed ServiceAccount landed before any decision record existed. This ADR
+documents what shipped and honestly records that the decision-before-code discipline was not
+followed here (see "Process note"). **Partially supersedes ADR-002 and ADR-013** (the "Tapir
+holds no write access to the shared identity provider" posture).
+
+**Keep-or-reverse, decided (2026-06-03).** After the retroactive recording, the maintainer
+weighed keep vs. reverse (drop the RBAC + invite flow, rely on Google OIDC only). **Decision:
+KEEP.** Deciding fact: not all intended Future-B users have / will use Google accounts, so Google
+OIDC alone *cannot* onboard them — the invite flow is therefore **load-bearing, not a redundant
+convenience**, and reversing it would leave some intended users with no onboarding path. The
+trust-surface cost (scoped `create` on `passwords` in `auth`) is accepted deliberately in
+exchange. The opposing argument (this contradicts ADR-015's "no write credential to shared infra"
+logic, applied there to secrets) was considered and outweighed *only* because the capability is
+genuinely necessary for real users here — it is **not** a precedent for widening the surface
+further. **Explicitly NOT chosen:** adding `delete` to the RBAC to close the orphan gap (that
+widens the surface in the wrong direction; accept the orphan at Future B instead — see open
+items).
+
+**Context.** Stage 1 onboarding needs a way for a friend to get a login. Two paths shipped:
+Google OIDC via Dex (no Tapir code — Dex handles it), and an **invite flow** where Tapir itself
+provisions a Dex **local-password** account. The invite flow is what this ADR is about.
+
+**What shipped (reconstructed from `internal/adapters/dex/dex.go`, `internal/web/invite.go`,
+migration `009_invitations`).**
+1. `tapir invite <email>` (host CLI) writes an `invitations` row: a 32-byte crypto-random,
+   single-use, 7-day-expiry `token` (the token is the capability), `email`, `used_at`.
+   The `invitations` table is deliberately **not** RLS/`user_id`-scoped — the invitee has no
+   user yet and the token itself is the secret. (Sound; documented in the migration.)
+2. Recipient visits `/invite/{token}` (public, no session), sets a password (validated *before*
+   the token is consumed, so a typo doesn't burn it), the token is claimed exactly once.
+3. Tapir bcrypt-hashes (cost 12) and **POSTs a `passwords.dex.coreos.com` Custom Resource into
+   the `auth` namespace** via the in-cluster Kubernetes API, authenticating as the `tapir`
+   ServiceAccount. TLS validated against the mounted cluster CA. Off-cluster (dev) it returns
+   `ErrNotInCluster` and degrades without consuming the invite.
+4. RBAC applied this deploy: the `tapir` ServiceAccount has **`create`+`get` on
+   `passwords.dex.coreos.com` in the `auth` namespace** (and only that).
+5. Dex (kubernetes storage) then serves local-password login for that email; Tapir's
+   registration gate creates the profile on first login.
+
+**Decision (as ratified now).** Accept the invite flow as built: Tapir *may* hold scoped write
+access to Dex's `passwords` resource in the `auth` namespace, for the purpose of invite-based
+local-account creation. This is a deliberate, bounded reversal of the prior "no write access to
+the shared identity provider" posture (ADR-002/013).
+
+**Why this is acceptable (the case for keeping it).**
+- The RBAC is **minimally scoped**: `create`+`get` on one resource type in one namespace, not
+  broad Dex/cluster write. Blast radius is bounded and inspectable.
+- It enables friend onboarding **without Google OAuth app verification** (ADR-008 deferred that),
+  which is genuinely useful for Future B — and necessary for users who won't use Google (the
+  deciding fact in the keep decision above).
+- The code is careful: validate-before-consume, single-use tokens, graceful off-cluster degrade,
+  sentinel errors mapped to clear messages, the base64-bcrypt storage gotcha handled.
+
+**Consequences / known gaps (the case to watch).**
+- **Reverses a load-bearing principle.** ADR-002 and ADR-013 leaned on "Tapir never writes to the
+  shared identity provider" for blast-radius minimisation. That is no longer true. Anyone reading
+  those ADRs without this one would be misled — hence the cross-references added to both.
+- **Create/delete asymmetry → orphaned Dex accounts.** Tapir now *creates* Dex Password CRs but
+  (per ADR-013) does not *delete* them on account deletion. A deleted Tapir user leaves an
+  orphaned Dex local-password account that can still authenticate (though it would hit the
+  registration gate with no profile). This widens the ADR-013 right-to-erasure gap. **Open —
+  accepted for Future B, revisit before Future C (do NOT add `delete` RBAC just to fix this).**
+- **`create` on a shared-namespace identity resource** is a meaningfully larger trust surface than
+  the rest of Tapir. If the `tapir` pod is compromised, the attacker can mint Dex local accounts.
+  Bounded by the namespace/resource scope, but real — worth a deliberate look before Future C.
+- Token custody for invites is in PG (`invitations.token`) in plaintext; single-use + 7-day TTL
+  bound the exposure, but a DB read yields live invite tokens until claimed/expired.
+
+**Process note (why this ADR is retroactive).** This capability was built and deployed by
+parallel agent sessions while the main planning thread was elsewhere, and shipped with no ADR —
+the first time in this project a guardrail-reversing change skipped the decision-before-code
+discipline. Recorded here not to rubber-stamp it but to restore the audit trail: the decision is
+now visible, its principle-reversal is named, and its open gaps (orphaned accounts, the larger
+trust surface) are tracked rather than buried in a release note. The maintainer ratified KEEP
+after the fact (see status block) on the deciding fact that some intended users cannot use Google
+OIDC.
+
+**Open items this ADR creates (tracked in infra):**
+- Confirm the RBAC really is `create`+`get` only (not broader) against the deployed manifest in
+  `infra` — a security claim currently resting on a sibling report, not a verified manifest read.
+- Accept the orphaned-Dex-account gap for Future B; revisit (orphan cleanup + the whole Dex-write
+  surface) before any Future C move. Do **not** add `delete` RBAC solely to fix the orphan.
+- Review the larger trust surface before any Future C move.
+
+---
+
+## ADR-018 — Make Tapir usable unprompted: in-process scheduled discovery, auto-summarize default, gate-clock reset
+
+**Status:** Accepted (2026-06-03)
+
+**Context.** The Stage-0 gate (ADR-016) measures whether the maintainer or a friend *returns and
+reads/acts* over weeks. But the system could not actually be used that way: discovery (`tapir
+run`) was **host-side manual**, so a newly onboarded user saw an empty list and had no reason to
+return. The gate was structurally unmeetable — not because the product failed, but because the
+*workflow* depended on the maintainer SSHing in to trigger each pass. The runner already has a
+`Loop(ctx, interval)` (run-on-start, then every tick) and per-user mode/dedup/backoff; what was
+missing was *invocation* — nothing called it for the deployed users on a schedule.
+
+**Decision.**
+1. **In-process scheduled discovery.** `tapir serve` launches a background goroutine that runs
+   discovery for **all users** on an interval (`TAPIR_DISCOVERY_INTERVAL`; unset/0 = off). It
+   enumerates users (un-RLS'd `user_identities`) and runs each user's pass inside `withUser`,
+   reusing the existing `runner` — not a new scheduler. Stateless timing; ctx-cancellable;
+   per-user failures isolated.
+2. **Auto-summarize default ON** for Future-B users (new registrations default `true`; existing
+   rows updated), so discovery both populates *and* summarizes — the list fills itself. The
+   per-user manual toggle remains.
+3. **Gate-clock reset.** The Stage-0 3–4 week window (ADR-016) **starts when this ships**, because
+   unprompted use was impossible before it. This is starting the clock when the experiment can
+   actually run, **not** a reset to dodge a failing gate (the prior window measured nothing —
+   there was no way to use the system unprompted). The 2026-07-01 check-in moves accordingly to
+   ~3–4 weeks after this deploys.
+
+**Why in-process and not a k8s CronJob.** Maintainer's call: simpler deploy (no second deployable),
+acceptable at Future-B scale (1–3 users). The CronJob's advantage — failure isolation between
+discovery and the web/read path — was weighed and traded away knowingly.
+
+**Consequences / constraints.**
+- **Discovery shares the web process's lifetime and egress.** A wedged discovery pass can degrade
+  the reading UI (the coupling a CronJob would have avoided). Accepted at this scale.
+- **SINGLE-REPLICA ASSUMPTION (load-bearing).** If `tapir serve` ever runs >1 replica, *every*
+  replica runs the discovery loop → every user fetched in parallel (429s + duplicate work). Tapir
+  must stay single-replica while in-process scheduling is enabled, or this is revisited (move to
+  CronJob, or add leader-election). Recorded so a future scale-up doesn't silently double-run.
+- **Makes ADR-014 item 2 (shared per-egress-IP rate gate) load-bearing.** In-process + auto +
+  multi-user drives all caption fetches through one pod egress, concurrent with click-path
+  Summarize. The build confirms/finishes the process-wide rate gate in the same slice; without it
+  the system self-inflicts 429s every cycle. **Auto-summarize ON is gated on the rate gate
+  existing** (fallback: ship discovery with auto OFF until it does).
+
+**Reversibility.** Disable via `TAPIR_DISCOVERY_INTERVAL=0` (reverts to manual `tapir run`).
+Moving to a CronJob later is a superseding ADR; the per-user `runner` is unchanged either way.
+Spec: `docs/specs/scheduled-discovery.md`.
+
+---
+
+## ADR-019 — Authentik owns invites; Tapir stops provisioning accounts
+
+**Status:** Accepted (2026-06-07). **Supersedes ADR-017** (Dex local-password invite
+provisioning).
+
+**Context:** infra ADR-0001 migrated the homelab IdP Dex→Authentik. Authentik provides
+first-class invite flows; the Dex local-password connector never consulted the Password
+CRs Tapir wrote (the defect that triggered the migration). Tapir-web's OIDC issuer now
+points at Authentik (infra ADR-0001 step 3).
+
+**Decision:** Tapir no longer provisions accounts. The Dex-password invite path is removed:
+`internal/adapters/dex`, the public `/invite/{token}` set-password UI (`internal/web/invite.go`),
+the `tapir invite` CLI (`cmd/tapir/invite.go`), the `InvitationStore`/`DexPasswordCreator`
+ports + `App.Invitations`/`App.Dex` wiring, the invite Templ pages, and the
+`tapir invite` Taskfile target. New users are invited via Authentik's flow, log into Tapir
+via OIDC, and are captured by Tapir's existing provider-agnostic `/register` (display name).
+Login + Google moved by config only (Authentik per-app issuer); the OIDC adapter is unchanged.
+
+**Consequences:** smaller Tapir blast surface — no writes to the shared identity provider, no
+configmap/CR access, the dedicated `passwords.dex.coreos.com` RBAC + ServiceAccount are
+removed (infra side, coupled change). The `invitations` table (migration 009) is left in
+place — migrations are append-only and the unused table is harmless; a future migration may
+drop it. The `*_DEX_*` config/identity names (`TAPIR_DEX_CLIENT_*`, `dex_subject`, the
+`DexAuth`/`oidc` package) are now misnomers; renaming is deferred (cosmetic, not behavioural).
+
+**Rejected:** keeping Tapir's `/invite` UI but calling Authentik's API on claim — couples
+Tapir to Authentik's admin API + a token for no real gain; Authentik's own invite flow is
+the supported path.
+
+---
+
+## ADR-020 — Recency-bounded auto-summarize + honest sparse-state surface
+
+**Status:** Accepted (2026-06-08). **Refines ADR-018** (auto-summarize) and **ADR-014**
+(per-IP caption rate gate).
+
+**Context.** Tapir is operational but sparse: at real subscription volume the maintainer's
+account holds ~283 discovered videos, ~15 summarized, ~256 behind the respected per-IP caption
+rate gate, ~12 no-captions. Two problems follow. (1) **Load:** ADR-018 auto-summarizes *every*
+unseen video, so a large back-catalogue re-drives the whole queue through the gate every cycle —
+self-inflicted 429s with no user value (nobody is waiting on a 6-month-old video). (2) **First
+contact:** a new user sees a mostly-empty feed with no moving parts and a UI that implied
+abundance/imminence ("fetching soon" ×256, "Run `tapir run`", "Summarize now"); the Stage-0 gate
+is *return usage*, and the experience died at the first visit. Source: a UX heuristic review
+(`docs/ux-review/UX-REVIEW-stage0-recency.md`).
+
+**Decision.**
+1. **Recency bound on auto-summarize.** In automatic mode the scheduler only summarizes videos
+   published within `TAPIR_AUTO_SUMMARIZE_WINDOW` (default ~7d). Older videos are still discovered
+   and listed but not auto-processed — they keep the manual "Summarize" affordance. An explicit
+   manual request bypasses the bound. `0` disables it (pre-recency behaviour). This bounds auto
+   *load*; it does not fetch harder — the gate (ADR-014) is untouched and the manual path still
+   serialises through it.
+2. **Honest sparse-state surface.** Copy is reframed to surface scarcity truthfully, never to
+   look fuller: "N ready · M in queue · K no captions" (not "fetching soon"); a one-line "captions
+   are fetched slowly on purpose" note; "Summarize" (not "Summarize now"); the empty-connected
+   state stops printing an impossible CLI command.
+3. **Feed IA = one list, noise-collapsed.** Summarized + recent un-summarized cards lead inline;
+   the older un-summarized back-catalogue collapses behind a single "Show N older videos"
+   disclosure; caption-less videos collapse to a one-line count instead of N dead cards. List is
+   ordered summarized-first, then `published_at DESC NULLS LAST`.
+
+**Consequences.** The auto path's per-cycle fetch volume is bounded by recent uploads, not the
+whole back-catalogue, so steady-state 429 pressure drops sharply. Older videos become explicitly
+on-demand — a deliberate honesty trade (the user chooses to spend a scarce fetch on old content).
+The single-replica assumption (ADR-018) is unchanged.
+
+**Deliberately NOT done (premature until the Stage-0 loop is validated).** Return-nudges
+(digest email / push) — a nudge contaminates the *unprompted*-return signal the gate measures
+(ADR-016); building it now poisons the experiment. Also deferred: full-text search, channel
+facets, read/unread, saved views — all need summary abundance to matter.
+
+**Reversibility.** `TAPIR_AUTO_SUMMARIZE_WINDOW=0` restores summarize-every-unseen; the feed
+collapse keys off the same window (`App.RecencyWindow=0` → everything inline).
+
+---
+
 ## Rejected alternatives
 
 Approaches considered during the 2026-06-02 planning + grill session and **deliberately not
@@ -308,6 +761,11 @@ maps to the ADR that settles it.
 | Audio-download + Whisper STT in the core path | ToS-grey, breakage-prone (yt-dlp), contends for koala GPU with the JEPA PoC; captions alone test the core hypothesis | ADR-007 |
 | Building multi-tenant SaaS / Google OAuth verification now | "Real users soon" was lowered to Future B; SaaS machinery before the Stage 0 self-use gate is the primary documented anti-goal | ADR-008, VISION |
 | Delegating the S5 reuse spike to an agent swarm | A 1-hour sequential read-and-judge with a single coupled conclusion; orchestration overhead exceeds the work, and it's Diamond-1 judgment the maintainer wanted to own | (process note) |
+| Vault-write SA for per-user OAuth tokens (ESO as runtime write path) | ESO syncs vault→cluster at deploy time, not a runtime write API; a write-SA widens blast radius to shared infra to store app row-data | ADR-015, infra#88 |
+| Supabase for per-user credential storage | Adds a second datastore for a few encrypted strings PG18 already holds; reopens ADR-002 | ADR-015, infra#88 |
+| Feedback-based Stage 0 gate (friends saying it's useful) | Politeness bias makes asked-for feedback the least reliable signal; return-usage is the real test | ADR-016 |
+| Reverse the Dex-write invite flow (Google OIDC only) | Some intended Future-B users won't use Google; OIDC-only leaves them with no onboarding path — invite flow is load-bearing | ADR-017 |
+| k8s CronJob for scheduled discovery (vs in-process) | At Future-B scale the in-process scheduler is simpler to deploy; CronJob's failure-isolation benefit was weighed and traded away knowingly (revisit if >1 replica or load grows) | ADR-018 |
 
 If a future case genuinely reopens one of these, that's a new ADR superseding the relevant one —
 not a silent reversal.

README.md

@@ -37,7 +37,7 @@ model, behavior specs) remain the source of intent.
 
 ## Running the Stage-0 demo
 
-Tapir runs on **your own** YouTube account: authorize once, then run the
+Tapir runs on your YouTube account(s): authorize once, then run the
 watch→summarize→deliver loop. All configuration is via `TAPIR_*` environment
 variables — copy [`.env.example`](.env.example) to `.env` and fill it in (no
 secrets are committed; at demo time source them from op, e.g. `op run -- ...`).
@@ -56,7 +56,7 @@ go build -o bin/tapir ./cmd/tapir
 ./bin/tapir auth
 
 # 3. run: detect new videos across your subscriptions, summarize, deliver to the
-#    store. Unset TAPIR_POLL_INTERVAL = single pass; set it (e.g. 15m) to loop.
+#    store. Single pass; set TAPIR_DISCOVERY_INTERVAL (e.g. 2h) for the serve loop.
 ./bin/tapir run
 ```
 
@@ -68,6 +68,18 @@ URI matches `TAPIR_OAUTH_REDIRECT_ADDR`. The summarizer model
 (`TAPIR_SUMMARIZER_MODEL`, default `koala/phi4-mini`) is overridable; pick the
 final alias when the gateway is reachable (see `docs/homelab-integration.md`).
 
+### Web surface (`tapir serve`)
+
+`tapir serve` starts the HTMX+Templ web UI on `:8080`. Users log in via Dex OIDC (local
+password or Google); a new Dex subject is routed to `/register` to create a Tapir account.
+Stage 1 is multi-user: each user connects their own YouTube account from the browser and
+manages their own summaries under DB-enforced RLS isolation. When `TAPIR_DISCOVERY_INTERVAL`
+is set (e.g. `2h`), the serve process runs a scheduled discovery pass for every registered
+user automatically — no CronJob required. In auto mode only videos published within
+`TAPIR_AUTO_SUMMARIZE_WINDOW` (default ~7d, ADR-020) are summarised automatically; older videos
+are listed and summarised on demand, so a large back-catalogue doesn't keep re-driving the
+caption rate gate. See `docs/homelab-integration.md` for the full config reference.
+
 ### Headless on koala
 
 koala has no browser and no interactive `op` session, so the two interactive

VISION.md

@@ -44,50 +44,74 @@ fallback — their key, their choice.
 
 ## Who it is for
 
-- **Now (the first customer):** the maintainer — one person, their own subscriptions,
+- **Now (the first customers):** the maintainer and a small number of known, trusted
-  summaries delivered to their own store and brain.
+  friends — each with their own account, isolated data, optional BYO-AI. The maintainer is
-- **Soon (Future B):** a small number of known, trusted users (friends / beta) — each with
+  the first customer; friendly users provide the earliest real-world signal.
-  their own account, isolated data, optional BYO-AI.
 - **Maybe (Future C, explicitly not built yet):** a public multi-tenant service. Deferred
-  until there is evidence of sustained personal use **and** real demand. Building for C
+  until there is evidence of sustained use **and** real demand. Building for C before that
-  before that evidence is a known anti-goal.
+  evidence is a known anti-goal.
 
 ## Definition of Success
 
 Success is staged. Each stage has a single, falsifiable headline test. We do not advance
 to the next stage's ambition until the current stage's test passes.
 
-### Stage 0 — Useful to me (the gate)
+### Stage 0 — Useful to me or a friend (the gate)
 
-> **Headline test:** For four consecutive weeks, the maintainer reads Tapir-produced
+> **Headline test:** Over a 3–4 week window, *either* the maintainer *or* at least one
-> summaries for their own subscriptions at least weekly, and at least once acts on a
+> onboarded friend returns to Tapir and reads/acts on summaries in **≥2 separate weeks**.
-> summary (watches / skips / saves a video *because of* the summary).
+> The test is *return usage* (behavioural), not stated approval. The ideal signal is an
+> **unprompted** return (organic, not because the maintainer nudged them) — but see the
+> measurement note below: we currently cannot distinguish prompted from organic returns, so
+> in practice we count all returns and read the result with that caveat.
 
-- Captions-first summarization works end-to-end for the maintainer's real subscriptions.
+- Captions-first summarization works end-to-end for real subscriptions (the maintainer's
-- Summaries land in the maintainer's store and (optionally) brain.
+  and onboarded friends').
+- Summaries land in each user's own store and (optionally) brain.
 - Local-first AI produces summaries of acceptable quality without manual intervention
   most of the time.
-- **This is the gate.** Multi-user, BYO-AI-for-others, and any SaaS ambition stay deferred
+- **Why behavioural, not feedback.** Friend *feedback* is gathered and genuinely valuable —
-  until Stage 0 holds. (Ties to the 2026-07-01 self-use check-in.)
+  but it is **not** the gate. Asked-for feedback from friendly users is the least reliable
+  signal in product development (politeness bias); whether they *come back* is the thing we
+  actually care about. So the gate measures returns, not nice words.
+- **Measurement note — "unprompted" is an ideal we can't yet measure.** Whether a return was
+  organic or prompted by a nudge is not captured by any data Tapir holds (it's context only
+  the maintainer has). Rather than waive the standard, we name the gap: *unprompted* return
+  is the signal we genuinely want; *returns* (prompted or not) is what the data can show. A
+  return that needed a nudge is a weaker signal than one that didn't, and the result is read
+  with that in mind. If distinguishing them ever matters enough, the maintainer tracks nudges
+  manually or a future build records prompt events — neither is in scope now.
+- **Why "me OR a friend".** This replaces the original "useful to *me*, specifically" gate
+  (2026-06-03 decision, recorded in DECISIONS.md ADR-016). Getting signal from friendly
+  users is valuable enough to count — but the bar stays behavioural so it can't be cleared
+  by a polite reaction. (Ties to the 2026-07-01 check-in.)
+- **How it's measured.** Return usage is read from two sources: `summary_actions` (timestamped
+  watch/skip/save per user) answers "acted in ≥2 distinct weeks"; an append-only login-events
+  table (see infra/Tapir build) answers "returned/read in ≥2 distinct weeks" even without an
+  action click — the honest signal for a *reading* product. Login events accrue only from their
+  deploy date onward, so the gate window's data begins then.
+- **Gate-clock reset (ADR-018).** The 3–4 week window starts when in-process scheduled discovery
+  + auto-summarize ship — before that, unprompted use was impossible, so the prior window
+  measured nothing (this is starting the clock when the experiment can actually run, not a reset
+  to dodge a failing gate). The 2026-07-01 check-in referenced above moves accordingly to ~3–4
+  weeks after this deploys. See DECISIONS.md ADR-018.
+- **This is the gate.** Hardening (Stage 1) and any SaaS ambition stay deferred until this
+  behavioural signal exists. Note: multi-user machinery was deliberately built *ahead* of
+  this gate (ADR-012) with isolation enforced — that was an explicit, recorded call, not a
+  sign the gate had passed. The gate is about *evidence of use*, which is still open.
 
-### Stage 1 — Useful to a few (Future B)
+### Stage 1 — Trustworthy at rest (hardening, Future B)
-
-> **Headline test:** At least one trusted user other than the maintainer connects their
-> own account and, within their first month, keeps using it (returns to read summaries in
-> ≥2 separate weeks) without the maintainer hand-holding each summary.
-
-- Multiple users, each with isolated accounts, credentials, and summaries.
-- A new user can self-connect a YouTube/Vimeo account and get summaries with no code change.
-- Optional BYO-AI works per-user.
-- No cross-user data leakage — demonstrable, not assumed.
-
-### Stage 2 — Trustworthy at rest (hardening, still Future B)
 
 > **Headline test:** Credentials (OAuth tokens, BYO-AI keys) are encrypted at rest via the
 > homelab's existing secrets convention; a documented, rehearsed recovery path exists; and
 > a deliberate isolation test (user A cannot read user B's data) passes in CI or a
 > documented manual drill.
 
+- Per-user data isolation is enforced and tested (delivered early via ADR-012 RLS).
+- Per-user credentials are encrypted at rest (ADR-015 envelope encryption; build in infra#89).
+- A new user can self-connect a YouTube/Vimeo account and get summaries with no code change.
+- Optional BYO-AI works per-user.
+
 ### Non-goals (current)
 
 - Public sign-up / billing / a marketing surface.
@@ -98,7 +122,11 @@ to the next stage's ambition until the current stage's test passes.
 
 ## How we will know we are drifting
 
-- We are building Stage 1+ machinery before the Stage 0 gate has passed.
+- We declare the Stage 0 gate "passed" on the strength of polite feedback rather than
+  behavioural return-usage (the politeness-bias trap the gate is designed to resist).
+- We build Stage 1 hardening or Future C machinery while the Stage 0 use-evidence is still
+  absent. (Multi-user machinery already shipped ahead of the gate via ADR-012 — a recorded,
+  deliberate exception, not a precedent for more.)
 - A user's content reaches a third-party model without that user's explicit, per-user opt-in.
 - "Brain ingestion" starts dictating the architecture instead of being one sink behind an
   interface.

cmd/tapir/discovery.go

@@ -0,0 +1,44 @@
+package main
+
+import (
+	"context"
+	"log/slog"
+	"sync"
+
+	"gitea.d-ma.be/mathias/tapir/internal/runner"
+)
+
+// discoveryRunner runs one user's discovery pass.
+type discoveryRunner func(ctx context.Context, userID string) (runner.Stats, error)
+
+// serialize wraps run so calls never overlap: every discovery pass — scheduled
+// or connect-triggered (#6) — acquires the same lock, preserving the
+// one-fetcher-at-a-time invariant the scheduler relies on (ADR-018, the
+// single-replica assumption). Locking is per-user, so a connect-triggered pass
+// interleaves between the scheduler's users instead of waiting for a whole pass.
+func serialize(mu *sync.Mutex, run discoveryRunner) discoveryRunner {
+	return func(ctx context.Context, userID string) (runner.Stats, error) {
+		mu.Lock()
+		defer mu.Unlock()
+		return run(ctx, userID)
+	}
+}
+
+// discoveryTrigger fires an out-of-band discovery pass for one user without
+// blocking the caller (the connect HTTP handler). The pass runs on the server's
+// long-lived ctx — not the request ctx — so it survives the post-connect
+// redirect. run is the serialized runner, so a trigger never overlaps the
+// scheduler. Satisfies web.DiscoveryTrigger.
+type discoveryTrigger struct {
+	ctx context.Context
+	run discoveryRunner
+	log *slog.Logger
+}
+
+func (t *discoveryTrigger) Enqueue(userID string) {
+	go func() {
+		if _, err := t.run(t.ctx, userID); err != nil {
+			t.log.Warn("discovery: connect-triggered pass had errors", "user", userID, "err", err)
+		}
+	}()
+}

cmd/tapir/discovery_test.go

@@ -0,0 +1,62 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"sync"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/runner"
+)
+
+// serialize must guarantee at most one discovery pass runs at a time, so a
+// connect-triggered pass never fetches concurrently with the scheduler.
+func TestSerializeRunsOneAtATime(t *testing.T) {
+	var active, maxActive int32
+	run := func(_ context.Context, _ string) (runner.Stats, error) {
+		n := atomic.AddInt32(&active, 1)
+		for { // record the high-water mark of concurrent runs
+			m := atomic.LoadInt32(&maxActive)
+			if n <= m || atomic.CompareAndSwapInt32(&maxActive, m, n) {
+				break
+			}
+		}
+		time.Sleep(2 * time.Millisecond)
+		atomic.AddInt32(&active, -1)
+		return runner.Stats{}, nil
+	}
+
+	s := serialize(&sync.Mutex{}, run)
+	var wg sync.WaitGroup
+	for i := 0; i < 20; i++ {
+		wg.Add(1)
+		go func(i int) { defer wg.Done(); _, _ = s(context.Background(), fmt.Sprintf("u%d", i)) }(i)
+	}
+	wg.Wait()
+
+	require.Equal(t, int32(1), atomic.LoadInt32(&maxActive),
+		"serialize must run at most one pass at a time")
+}
+
+// Enqueue runs the user's pass out-of-band (non-blocking) on the trigger's ctx.
+func TestDiscoveryTriggerEnqueueRunsUser(t *testing.T) {
+	done := make(chan string, 1)
+	run := func(_ context.Context, userID string) (runner.Stats, error) {
+		done <- userID
+		return runner.Stats{}, nil
+	}
+	tr := &discoveryTrigger{ctx: context.Background(), run: run, log: quietLog()}
+
+	tr.Enqueue("u1")
+
+	select {
+	case got := <-done:
+		require.Equal(t, "u1", got)
+	case <-time.After(2 * time.Second):
+		t.Fatal("Enqueue did not run the user's pass")
+	}
+}

cmd/tapir/main.go

@@ -20,17 +20,15 @@ import (
 	"net/http"
 	"os"
 	"os/signal"
+	"sync"
 	"time"
 
-	"gitea.d-ma.be/mathias/tapir/internal/adapters/llm"
 	"gitea.d-ma.be/mathias/tapir/internal/adapters/secrets"
 	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
-	"gitea.d-ma.be/mathias/tapir/internal/adapters/summarizer"
 	"gitea.d-ma.be/mathias/tapir/internal/adapters/youtube"
 	"gitea.d-ma.be/mathias/tapir/internal/auth"
 	"gitea.d-ma.be/mathias/tapir/internal/config"
 	"gitea.d-ma.be/mathias/tapir/internal/runner"
-	"gitea.d-ma.be/mathias/tapir/internal/usecase"
 	"gitea.d-ma.be/mathias/tapir/internal/web"
 	"gitea.d-ma.be/mathias/tapir/internal/web/oidc"
 )
@@ -57,6 +55,8 @@ func main() {
 		err = cmdRun(ctx, log)
 	case "serve":
 		err = cmdServe(ctx, log)
+	case "report":
+		err = runReport(ctx, os.Args[2:])
 	default:
 		usage()
 		os.Exit(2)
@@ -77,6 +77,7 @@ usage:
   tapir serve             run the web UI (read summaries, record watch/skip/save)
   tapir list [-limit N]   list stored summaries, recent first
   tapir show <video-id>   show one summary in full
+  tapir report            Stage-0 usage gate: per-user distinct active weeks
 
 configuration is via TAPIR_* environment variables (see .env.example).
 `)
@@ -120,27 +121,26 @@ func cmdRun(ctx context.Context, log *slog.Logger) error {
 	}
 	defer st.Close()
 
-	secretStore := secrets.NewFileStore(cfg.SecretsFile)
+	// Same wiring the web serve path uses (buildProcessor). ValidateForRun above
-	src := youtube.New(youtube.Config{
+	// already required the engine's inputs, so a nil here is a genuine config gap.
-		ClientID:           cfg.YTClientID,
+	engine, err := buildProcessor(cfg, st)
-		ClientSecret:       cfg.YTClientSecret,
+	if err != nil {
-		TokenSecretRef:     cfg.YTTokenRef,
+		return err
-		PreferredLanguages: []string{"en"},
-	}, secretStore)
-
-	// Local Primary only; no BYO fallback for the demo (fallback nil).
-	primary := summarizer.Endpoint{
-		Client:   llm.New(cfg.GatewayURL, cfg.GatewayKey, cfg.SummarizerModel, cfg.SummarizerTimeout),
-		Provider: "local",
-		Model:    cfg.SummarizerModel,
 	}
-	sum := summarizer.New(primary, nil)
+	if engine == nil {
+		return fmt.Errorf("run: incomplete summarization config (gateway, youtube credentials, secrets file)")
+	}
+	// Process-wide caption-fetch rate gate (ADR-014 item 2): the batch path shares
+	// the same per-egress-IP limiter as the web click-path.
+	youtube.SetFetchRate(cfg.FetchRate)
 
-	engine := usecase.NewEngine(src, sum, st)
+	r := runner.New(engine.Source, st, engine, cfg.UserID, log,
-	r := runner.New(src, st, engine, cfg.UserID, log)
+		runner.WithBackoff(cfg.FetchBackoff),
+		runner.WithAutoWindow(cfg.AutoSummarizeWindow))
 
 	log.Info("starting run", "user", cfg.UserID, "model", cfg.SummarizerModel,
-		"gateway", cfg.GatewayURL, "poll_interval", cfg.PollInterval)
+		"gateway", cfg.GatewayURL, "poll_interval", cfg.PollInterval, "fetch_backoff", cfg.FetchBackoff,
+		"fetch_rate", cfg.FetchRate, "auto_window", cfg.AutoSummarizeWindow)
 	return r.Loop(ctx, cfg.PollInterval)
 }
 
@@ -164,6 +164,11 @@ func cmdServe(ctx context.Context, log *slog.Logger) error {
 	}
 	defer st.Close()
 
+	// Process-wide caption-fetch rate gate (ADR-014 item 2): the web click-path
+	// and the scheduled-discovery runners share one per-egress-IP limiter so they
+	// cannot collectively trip 429s. Must be set before either path fetches.
+	youtube.SetFetchRate(cfg.FetchRate)
+
 	// Auth seam (handlers depend on web.Auth only). With Dex configured
 	// (TAPIR_OIDC_ISSUER set) serve uses real OIDC login — any Dex subject may
 	// authenticate, then registers a tapir user (ADR-012); otherwise it falls
@@ -189,7 +194,11 @@ func cmdServe(ctx context.Context, log *slog.Logger) error {
 	// The file-backed SecretStore is shared by the connect flow (writes tokens)
 	// and account management (deletes them on disconnect / delete-account).
 	secretStore := secrets.NewFileStore(cfg.SecretsFile)
-	app := &web.App{Store: st, Identity: st, Auth: authn, Secrets: secretStore, Log: log}
+	app := &web.App{Store: st, Identity: st, Auth: authn, Secrets: secretStore, Log: log, RecencyWindow: cfg.AutoSummarizeWindow}
+
+	// User onboarding is handled by the IdP (Authentik invite flow), not Tapir —
+	// the Dex local-password provisioning path was removed (ADR-019). An
+	// authenticated subject with no Tapir user is routed to /register.
 
 	// Web-initiated YouTube connect (ADR-006). Mounted only when the OAuth client
 	// credentials are present; the refresh token persists through the SecretStore
@@ -204,6 +213,51 @@ func cmdServe(ctx context.Context, log *slog.Logger) error {
 		log.Info("web youtube connect enabled", "redirect", cfg.YTConnectRedirectURL)
 	}
 
+	// Immediate summarization for the web "Summarize" button. When the engine can
+	// be built (gateway + YouTube credentials + secrets present), a click runs the
+	// summary now in the background; otherwise the button stays queue-only and the
+	// next `tapir run` does the work (buildProcessor returns nil — never an error).
+	engine, err := buildProcessor(cfg, st)
+	if err != nil {
+		return err
+	}
+	if engine != nil {
+		app.Processor = &engineProcessor{engine: engine, store: st}
+		log.Info("web immediate summarization enabled", "model", cfg.SummarizerModel)
+	} else {
+		log.Info("web summarization is queue-only (incomplete engine config)")
+	}
+
+	// In-process scheduled discovery (ADR-018): when enabled, a background
+	// goroutine runs a discovery pass for ALL users on TAPIR_DISCOVERY_INTERVAL,
+	// reusing the per-user runner.Runner. Cancelled by the same ctx as the server.
+	//
+	// SINGLE-REPLICA ASSUMPTION (load-bearing): this loop lives in the web process.
+	// Running serve at >1 replica would make every replica fetch every user in
+	// parallel — duplicate work and self-inflicted 429s. replicas: 1 is required in
+	// the deployment manifest; scaling up needs a CronJob or leader election first.
+	if cfg.DiscoveryInterval > 0 {
+		log.Info("scheduled discovery enabled", "interval", cfg.DiscoveryInterval, "fetch_rate", cfg.FetchRate)
+		log.Warn("scheduled discovery assumes a SINGLE replica — running serve at >1 replica double-runs discovery (ADR-018)")
+		rawRunUser := func(ctx context.Context, userID string) (runner.Stats, error) {
+			r, err := buildUserRunner(cfg, st, secretStore, userID, log)
+			if err != nil {
+				return runner.Stats{}, err
+			}
+			return r.RunOnce(ctx)
+		}
+		// One lock shared by the scheduler and connect-triggered passes (#6) so
+		// they never fetch concurrently — the single-fetcher invariant (ADR-018).
+		runUser := serialize(&sync.Mutex{}, rawRunUser)
+		if app.Connect != nil {
+			app.Connect.Discovery = &discoveryTrigger{ctx: ctx, run: runUser, log: log}
+			log.Info("connect-triggered discovery enabled")
+		}
+		go runScheduler(ctx, cfg.DiscoveryInterval, st, runUser, log)
+	} else {
+		log.Info("scheduled discovery disabled (TAPIR_DISCOVERY_INTERVAL unset or 0)")
+	}
+
 	srv := &http.Server{
 		Addr:              cfg.HTTPAddr,
 		Handler:           app.Router(),

cmd/tapir/processor.go

@@ -0,0 +1,86 @@
+package main
+
+import (
+	"context"
+	"fmt"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/llm"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/secrets"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/summarizer"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/youtube"
+	"gitea.d-ma.be/mathias/tapir/internal/config"
+	"gitea.d-ma.be/mathias/tapir/internal/domain"
+	"gitea.d-ma.be/mathias/tapir/internal/usecase"
+)
+
+// buildProcessor wires the summarization engine — YouTube source (captions-first),
+// AI-router summarizer, store sink — shared by `tapir run` and the web
+// "Summarize now" path so the wiring lives in one place. It returns (nil, nil) —
+// not an error — when the config cannot support live summarization (no gateway
+// URL, no YouTube client credentials, or no secrets file). That nil is the
+// queue-only fallback: the web UI keeps working (the button just queues) and
+// `tapir run` reports the gap via its own ValidateForRun. Missing engine config
+// is never an error here.
+func buildProcessor(cfg config.Config, st *store.Store) (*usecase.Engine, error) {
+	if cfg.GatewayURL == "" || cfg.YTClientID == "" || cfg.YTClientSecret == "" || cfg.SecretsFile == "" {
+		return nil, nil
+	}
+
+	secretStore := secrets.NewFileStore(cfg.SecretsFile)
+	src := youtube.New(youtube.Config{
+		ClientID:           cfg.YTClientID,
+		ClientSecret:       cfg.YTClientSecret,
+		TokenSecretRef:     cfg.YTTokenRef,
+		PreferredLanguages: []string{"en"},
+	}, secretStore)
+
+	// Local Primary only; no BYO fallback for the demo (fallback nil).
+	primary := summarizer.Endpoint{
+		Client:   llm.New(cfg.GatewayURL, cfg.GatewayKey, cfg.SummarizerModel, cfg.SummarizerTimeout),
+		Provider: "local",
+		Model:    cfg.SummarizerModel,
+	}
+	sum := summarizer.New(primary, nil)
+
+	return usecase.NewEngine(src, sum, st), nil
+}
+
+// engineProcessor adapts the engine (which works in terms of a domain.Video) to
+// the web.Processor port (which works in terms of a stored video id): it loads the
+// video row, runs the engine, and — on a produced summary — clears the manual
+// queue flag, mirroring the runner so the video is not re-summarized on the next
+// `tapir run` and the UI drops the "Queued" chip. A skip (no transcript) leaves
+// the flag set so a later run can retry.
+type engineProcessor struct {
+	engine *usecase.Engine
+	store  *store.Store
+}
+
+func (p *engineProcessor) ProcessVideo(ctx context.Context, userID, videoID string) error {
+	row, err := p.store.GetVideoRow(ctx, userID, videoID)
+	if err != nil {
+		return fmt.Errorf("load video %q: %w", videoID, err)
+	}
+
+	v := domain.Video{
+		ID:              row.VideoID,
+		UserID:          userID,
+		Provider:        domain.Provider(row.Channel),
+		ProviderVideoID: row.ProviderVideoID,
+		Title:           row.Title,
+		URL:             row.URL,
+		PublishedAt:     row.PublishedAt,
+	}
+
+	res, err := p.engine.ProcessNewVideo(ctx, v)
+	if err != nil {
+		return fmt.Errorf("process video %q: %w", videoID, err)
+	}
+	if res.Summary != nil {
+		if err := p.store.ClearSummarizeRequested(ctx, userID, videoID); err != nil {
+			return fmt.Errorf("clear summarize flag %q: %w", videoID, err)
+		}
+	}
+	return nil
+}

cmd/tapir/processor_test.go

@@ -0,0 +1,47 @@
+package main
+
+import (
+	"testing"
+
+	"gitea.d-ma.be/mathias/tapir/internal/config"
+)
+
+// TestBuildProcessorNilOnIncompleteConfig asserts the queue-only fallback: when a
+// required input is missing, buildProcessor returns (nil, nil) — never an error —
+// so the web UI degrades to queue-only instead of failing to start.
+func TestBuildProcessorNilOnIncompleteConfig(t *testing.T) {
+	// A complete config (the fields buildProcessor gates on). The store is nil:
+	// buildProcessor must not touch it on the incomplete paths, and the complete
+	// path only stores the pointer (no connection), so nil is fine for this test.
+	complete := config.Config{
+		GatewayURL:     "http://gw/v1",
+		YTClientID:     "id",
+		YTClientSecret: "secret",
+		SecretsFile:    "/tmp/secrets.json",
+	}
+
+	tests := []struct {
+		name    string
+		mutate  func(config.Config) config.Config
+		wantNil bool
+	}{
+		{"complete", func(c config.Config) config.Config { return c }, false},
+		{"no gateway url", func(c config.Config) config.Config { c.GatewayURL = ""; return c }, true},
+		{"no yt client id", func(c config.Config) config.Config { c.YTClientID = ""; return c }, true},
+		{"no yt client secret", func(c config.Config) config.Config { c.YTClientSecret = ""; return c }, true},
+		{"no secrets file", func(c config.Config) config.Config { c.SecretsFile = ""; return c }, true},
+		{"empty config", func(config.Config) config.Config { return config.Config{} }, true},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			engine, err := buildProcessor(tt.mutate(complete), nil)
+			if err != nil {
+				t.Fatalf("buildProcessor returned an error, want nil: %v", err)
+			}
+			if (engine == nil) != tt.wantNil {
+				t.Fatalf("engine == nil is %v, want %v", engine == nil, tt.wantNil)
+			}
+		})
+	}
+}

cmd/tapir/report.go

@@ -0,0 +1,67 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"os"
+	"text/tabwriter"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+)
+
+// gateThreshold is the Stage-0 gate (VISION/ADR-016): usage in >= 2 distinct
+// weeks. The gate passes when any user reaches it.
+const gateThreshold = 2
+
+// runReport prints the Stage-0 usage gate: per-user distinct active weeks (reads
+// UNION acts) and the pass/fail verdict. Read-only, cross-user — needs only
+// TAPIR_DB_DSN (not TAPIR_USER_ID; the report enumerates all users itself).
+func runReport(ctx context.Context, _ []string) error {
+	dsn := os.Getenv(envDSN)
+	if dsn == "" {
+		return fmt.Errorf("%s is required", envDSN)
+	}
+	s, err := store.New(ctx, dsn)
+	if err != nil {
+		return err
+	}
+	defer s.Close()
+
+	rows, err := s.ActiveWeeks(ctx)
+	if err != nil {
+		return err
+	}
+	return formatReport(os.Stdout, rows)
+}
+
+// formatReport renders the per-user week counts and the gate verdict. Pure: no DB,
+// no env — so the layout and verdict logic are unit-testable without Postgres.
+func formatReport(w io.Writer, rows []store.UserActiveWeeks) error {
+	if len(rows) == 0 {
+		_, err := fmt.Fprintln(w, "no users yet")
+		return err
+	}
+
+	tw := tabwriter.NewWriter(w, 0, 4, 2, ' ', 0)
+	_, _ = fmt.Fprintln(tw, "USER\tNAME\tACTIVE_WEEKS\tGATE")
+	passed := false
+	for _, r := range rows {
+		gate := "-"
+		if r.ActiveWeeks >= gateThreshold {
+			gate = "PASS"
+			passed = true
+		}
+		_, _ = fmt.Fprintf(tw, "%s\t%s\t%d\t%s\n", r.UserID, orDash(r.DisplayName), r.ActiveWeeks, gate)
+	}
+	if err := tw.Flush(); err != nil {
+		return err
+	}
+
+	verdict := fmt.Sprintf("\nGate (usage in >= %d distinct weeks): NOT YET MET\n", gateThreshold)
+	if passed {
+		verdict = fmt.Sprintf("\nGate (usage in >= %d distinct weeks): PASSED\n", gateThreshold)
+	}
+	_, err := fmt.Fprint(w, verdict)
+	return err
+}

cmd/tapir/report_test.go

@@ -0,0 +1,45 @@
+package main
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+)
+
+func TestFormatReportColumnsAndGatePass(t *testing.T) {
+	rows := []store.UserActiveWeeks{
+		{UserID: "user-a", DisplayName: "Ada", ActiveWeeks: 3},
+		{UserID: "user-b", DisplayName: "", ActiveWeeks: 1},
+	}
+
+	var b strings.Builder
+	require.NoError(t, formatReport(&b, rows))
+	out := b.String()
+
+	require.Contains(t, out, "USER")
+	require.Contains(t, out, "ACTIVE_WEEKS")
+	require.Contains(t, out, "Ada")
+	require.Contains(t, out, "PASSED", "a user at >= 2 weeks passes the gate")
+
+	// The >=2 user is marked PASS; the 1-week user is not.
+	require.Contains(t, lineContaining(t, out, "user-a"), "PASS")
+	require.NotContains(t, lineContaining(t, out, "user-b"), "PASS")
+	require.Contains(t, lineContaining(t, out, "user-b"), "-", "no display name falls back to dash")
+}
+
+func TestFormatReportGateNotMet(t *testing.T) {
+	rows := []store.UserActiveWeeks{{UserID: "user-a", ActiveWeeks: 1}}
+
+	var b strings.Builder
+	require.NoError(t, formatReport(&b, rows))
+	require.Contains(t, b.String(), "NOT YET MET", "no user at >= 2 weeks fails the gate")
+}
+
+func TestFormatReportEmpty(t *testing.T) {
+	var b strings.Builder
+	require.NoError(t, formatReport(&b, nil))
+	require.Contains(t, b.String(), "no users yet")
+}

cmd/tapir/scheduler.go

@@ -0,0 +1,154 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"log/slog"
+	"time"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/llm"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/summarizer"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/youtube"
+	"gitea.d-ma.be/mathias/tapir/internal/config"
+	"gitea.d-ma.be/mathias/tapir/internal/ports"
+	"gitea.d-ma.be/mathias/tapir/internal/runner"
+	"gitea.d-ma.be/mathias/tapir/internal/usecase"
+	"gitea.d-ma.be/mathias/tapir/internal/web"
+)
+
+// buildUserRunner constructs a runner.Runner for one user, reusing the same
+// engine wiring as buildProcessor but bound to that user's own YouTube refresh
+// token (web.YouTubeTokenRef(userID)) — the Stage-1 per-tenant ref, not the
+// Stage-0 single ref. It returns an error (not nil) when the global config can't
+// support live summarization (gateway, YouTube client creds, secrets file), so
+// the scheduler can skip that user gracefully. A user who simply hasn't connected
+// YouTube yet builds fine here; their token ref fails to resolve at RunOnce time,
+// surfacing as a per-user error the scheduler logs and skips.
+func buildUserRunner(cfg config.Config, st *store.Store, secretStore ports.SecretStore, userID string, log *slog.Logger) (*runner.Runner, error) {
+	if cfg.GatewayURL == "" || cfg.YTClientID == "" || cfg.YTClientSecret == "" || cfg.SecretsFile == "" {
+		return nil, fmt.Errorf("buildUserRunner: incomplete summarization config (gateway, youtube credentials, secrets file)")
+	}
+
+	src := youtube.New(youtube.Config{
+		ClientID:           cfg.YTClientID,
+		ClientSecret:       cfg.YTClientSecret,
+		TokenSecretRef:     web.YouTubeTokenRef(userID),
+		PreferredLanguages: []string{"en"},
+	}, secretStore)
+
+	primary := summarizer.Endpoint{
+		Client:   llm.New(cfg.GatewayURL, cfg.GatewayKey, cfg.SummarizerModel, cfg.SummarizerTimeout),
+		Provider: "local",
+		Model:    cfg.SummarizerModel,
+	}
+	engine := usecase.NewEngine(src, summarizer.New(primary, nil), st)
+
+	return runner.New(src, st, engine, userID, log,
+		runner.WithBackoff(cfg.FetchBackoff),
+		runner.WithAutoWindow(cfg.AutoSummarizeWindow)), nil
+}
+
+// userLister enumerates every registered user and reports a user's video
+// connections. *store.Store satisfies it via ListAllUsers + ConnectionsForUser.
+// A small local interface keeps the scheduler testable with a fake.
+type userLister interface {
+	ListAllUsers(ctx context.Context) ([]store.UserIdentity, error)
+	ConnectionsForUser(ctx context.Context, userID string) ([]store.Connection, error)
+}
+
+// runDiscoveryPass runs one discovery pass for every user. runUser performs a
+// single user's pass (production: build a runner and RunOnce). Per-user failures
+// — including a buildUserRunner error or a RunOnce error — are logged and skipped
+// so one bad user, channel, or video never aborts the others (ADR-018 failure
+// isolation). Returns the stats summed across users.
+func runDiscoveryPass(
+	ctx context.Context,
+	lister userLister,
+	runUser func(context.Context, string) (runner.Stats, error),
+	log *slog.Logger,
+) runner.Stats {
+	users, err := lister.ListAllUsers(ctx)
+	if err != nil {
+		log.Error("scheduler: list users failed", "err", err)
+		return runner.Stats{}
+	}
+
+	log.Info("scheduler: starting discovery pass", "users", len(users))
+	var total runner.Stats
+	for _, u := range users {
+		if ctx.Err() != nil {
+			break // shutting down: stop enumerating
+		}
+		// Skip users with no video connection. A discovery pass for them only
+		// attempts to resolve a token that was never minted, logging a spurious
+		// "ref not found" every tick (e.g. stale Dex-era orphan identities).
+		conns, err := lister.ConnectionsForUser(ctx, u.UserID)
+		if err != nil {
+			log.Warn("scheduler: list connections failed", "user", u.UserID, "err", err)
+			continue
+		}
+		if len(conns) == 0 {
+			log.Debug("scheduler: skipping user with no video connections", "user", u.UserID)
+			continue
+		}
+		stats, err := runUser(ctx, u.UserID)
+		total = sumStats(total, stats)
+		if err != nil {
+			log.Warn("scheduler: user discovery pass had errors", "user", u.UserID, "err", err)
+		}
+	}
+	log.Info("scheduler: pass complete",
+		"candidates", total.Candidates, "summarized", total.Summarized,
+		"skipped_seen", total.SkippedSeen, "skipped_no_text", total.SkippedNoText,
+		"skipped_manual", total.SkippedManual, "skipped_too_old", total.SkippedTooOld,
+		"skipped_rate_limited", total.SkippedRateLimited,
+		"channel_unavailable", total.ChannelUnavailable, "errors", total.Errors)
+	return total
+}
+
+// runScheduler runs a discovery pass on startup, then once every interval until
+// ctx is cancelled (pod SIGTERM exits the loop cleanly). A non-positive interval
+// disables scheduling entirely (no startup pass) so dev/tests never auto-fetch.
+// It reuses the existing runner.Runner via runUser — the only new behaviour over
+// runner.Loop is iterating all users per tick (ADR-018).
+func runScheduler(
+	ctx context.Context,
+	interval time.Duration,
+	lister userLister,
+	runUser func(context.Context, string) (runner.Stats, error),
+	log *slog.Logger,
+) {
+	if interval <= 0 {
+		return // disabled
+	}
+
+	runDiscoveryPass(ctx, lister, runUser, log)
+
+	ticker := time.NewTicker(interval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+			runDiscoveryPass(ctx, lister, runUser, log)
+		}
+	}
+}
+
+// sumStats adds two passes' stats field-wise, so runDiscoveryPass can report a
+// per-tick aggregate across all users.
+func sumStats(a, b runner.Stats) runner.Stats {
+	return runner.Stats{
+		Candidates:         a.Candidates + b.Candidates,
+		Summarized:         a.Summarized + b.Summarized,
+		SkippedSeen:        a.SkippedSeen + b.SkippedSeen,
+		SkippedNoText:      a.SkippedNoText + b.SkippedNoText,
+		SkippedManual:      a.SkippedManual + b.SkippedManual,
+		SkippedTooOld:      a.SkippedTooOld + b.SkippedTooOld,
+		SkippedRateLimited: a.SkippedRateLimited + b.SkippedRateLimited,
+		ChannelUnavailable: a.ChannelUnavailable + b.ChannelUnavailable,
+		Errors:             a.Errors + b.Errors,
+	}
+}

cmd/tapir/scheduler_test.go

@@ -0,0 +1,175 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"io"
+	"log/slog"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+	"gitea.d-ma.be/mathias/tapir/internal/runner"
+)
+
+func quietLog() *slog.Logger {
+	return slog.New(slog.NewTextHandler(io.Discard, nil))
+}
+
+// fakeLister returns a fixed user set (or an error) for the scheduler under test.
+type fakeLister struct {
+	users  []store.UserIdentity
+	err    error
+	noConn map[string]bool // users that have NOT connected a video source
+}
+
+func (f fakeLister) ListAllUsers(context.Context) ([]store.UserIdentity, error) {
+	return f.users, f.err
+}
+
+// ConnectionsForUser reports a single youtube connection for every user except
+// those in noConn, which return zero — the connection-less case the scheduler
+// must skip instead of running (and failing to resolve a token for).
+func (f fakeLister) ConnectionsForUser(_ context.Context, userID string) ([]store.Connection, error) {
+	if f.noConn[userID] {
+		return nil, nil
+	}
+	return []store.Connection{{Provider: "youtube"}}, nil
+}
+
+// countingRunUser records how many passes each user got, optionally failing for
+// specific users, under a mutex so it is safe across the scheduler goroutine.
+type countingRunUser struct {
+	mu      sync.Mutex
+	calls   map[string]int
+	failFor map[string]bool
+}
+
+func newCountingRunUser(failFor ...string) *countingRunUser {
+	c := &countingRunUser{calls: map[string]int{}, failFor: map[string]bool{}}
+	for _, u := range failFor {
+		c.failFor[u] = true
+	}
+	return c
+}
+
+func (c *countingRunUser) run(_ context.Context, userID string) (runner.Stats, error) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+	c.calls[userID]++
+	if c.failFor[userID] {
+		return runner.Stats{Errors: 1}, errors.New("boom")
+	}
+	return runner.Stats{Summarized: 1}, nil
+}
+
+func (c *countingRunUser) count(userID string) int {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+	return c.calls[userID]
+}
+
+func (c *countingRunUser) total() int {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+	n := 0
+	for _, v := range c.calls {
+		n += v
+	}
+	return n
+}
+
+func usersN(ids ...string) []store.UserIdentity {
+	out := make([]store.UserIdentity, len(ids))
+	for i, id := range ids {
+		out[i] = store.UserIdentity{UserID: id, DexSubject: "dex|" + id}
+	}
+	return out
+}
+
+func TestDiscoveryPassRunsEveryUserOnce(t *testing.T) {
+	lister := fakeLister{users: usersN("a", "b", "c")}
+	rc := newCountingRunUser()
+
+	stats := runDiscoveryPass(context.Background(), lister, rc.run, quietLog())
+
+	require.Equal(t, 1, rc.count("a"))
+	require.Equal(t, 1, rc.count("b"))
+	require.Equal(t, 1, rc.count("c"))
+	require.Equal(t, 3, stats.Summarized, "stats are summed across users")
+}
+
+func TestDiscoveryPassSkipsUsersWithoutConnections(t *testing.T) {
+	// b never connected a video source (e.g. a stale Dex-era orphan identity).
+	// It must be skipped silently — not run and logged as a token error every pass.
+	lister := fakeLister{users: usersN("a", "b", "c"), noConn: map[string]bool{"b": true}}
+	rc := newCountingRunUser()
+
+	stats := runDiscoveryPass(context.Background(), lister, rc.run, quietLog())
+
+	require.Equal(t, 1, rc.count("a"))
+	require.Equal(t, 0, rc.count("b"), "a user with no connection must be skipped, not run")
+	require.Equal(t, 1, rc.count("c"))
+	require.Equal(t, 2, stats.Summarized, "only connected users contribute")
+	require.Equal(t, 0, stats.Errors, "skipping is silent — no spurious error stat")
+}
+
+func TestDiscoveryPassOneUserFailureDoesNotStopOthers(t *testing.T) {
+	lister := fakeLister{users: usersN("a", "b", "c")}
+	rc := newCountingRunUser("b") // user b's pass errors
+
+	stats := runDiscoveryPass(context.Background(), lister, rc.run, quietLog())
+
+	require.Equal(t, 1, rc.count("a"))
+	require.Equal(t, 1, rc.count("b"))
+	require.Equal(t, 1, rc.count("c"), "a failing user must not abort the rest")
+	require.Equal(t, 2, stats.Summarized) // a + c
+	require.Equal(t, 1, stats.Errors)     // b
+}
+
+func TestDiscoveryPassListerErrorIsContained(t *testing.T) {
+	lister := fakeLister{err: errors.New("db down")}
+	rc := newCountingRunUser()
+
+	stats := runDiscoveryPass(context.Background(), lister, rc.run, quietLog())
+
+	require.Equal(t, 0, rc.total(), "no users enumerated → no passes")
+	require.Equal(t, runner.Stats{}, stats)
+}
+
+func TestSchedulerIntervalZeroDisablesEntirely(t *testing.T) {
+	lister := fakeLister{users: usersN("a", "b")}
+	rc := newCountingRunUser()
+
+	runScheduler(context.Background(), 0, lister, rc.run, quietLog())
+
+	require.Equal(t, 0, rc.total(), "interval 0 must not run even a startup pass")
+}
+
+func TestSchedulerRunsStartupPassThenStopsOnCancel(t *testing.T) {
+	lister := fakeLister{users: usersN("a", "b", "c")}
+	rc := newCountingRunUser()
+
+	ctx, cancel := context.WithCancel(context.Background())
+	done := make(chan struct{})
+	go func() {
+		// A long interval so only the startup pass runs before we cancel.
+		runScheduler(ctx, time.Hour, lister, rc.run, quietLog())
+		close(done)
+	}()
+
+	// The startup pass is synchronous at the top of runScheduler; once all three
+	// users have a pass it has completed and the loop is parked on the ticker.
+	require.Eventually(t, func() bool { return rc.total() == 3 }, time.Second, 5*time.Millisecond)
+
+	cancel()
+	select {
+	case <-done:
+	case <-time.After(time.Second):
+		t.Fatal("scheduler did not exit after ctx cancel")
+	}
+	require.Equal(t, 3, rc.total(), "no extra passes fired between startup and cancel")
+}

docs/architecture/architecture.md

@@ -45,7 +45,7 @@ adapter behind an interface (Clean Architecture ports & adapters).
 ```mermaid
 graph TB
     subgraph tapir["Tapir (Go)"]
-        http["HTTP server<br/>OAuth callbacks +<br/>user-facing API"]
+        http["tapir serve<br/>(HTMX+Templ web surface:<br/>read summaries, connect,<br/>account, summarize)"]
         watcher["Watcher<br/>detects new videos<br/>(WebSub + poll)"]
         engine["Summarization engine<br/>(use-case core)"]
         resolver["Transcript resolver<br/>(captions-first)"]
@@ -95,6 +95,163 @@ two codebases (ADR-003).
 
 ---
 
+## Web surface — `tapir serve` (Stage 1, ADR-011 → ADR-012)
+
+A later transport added over the **unchanged** engine/ports/sinks core (ADR-003): `tapir serve`
+is an HTMX+Templ reader/writer (`internal/web`) over the existing `store`. It added no business
+logic to the engine — it reads the store and, for one action, kicks the existing engine. ADR-011
+shipped it single-user; ADR-012 opened multi-user with DB-enforced (RLS) isolation.
+
+```mermaid
+graph TB
+    browser["Browser<br/>(Dex-authenticated user)"]
+    subgraph web["internal/web (tapir serve)"]
+        oidc["oidc<br/>Dex OIDC session<br/>(authenticate-only)"]
+        gate["registration gate<br/>new subject -> /register"]
+        pages["summary list + detail<br/>(read) + actions"]
+        connect["/oauth/youtube/callback<br/>per-user token connect"]
+        account["account<br/>(disconnect, delete)"]
+        summarize["Summarize button<br/>-> background goroutine"]
+    end
+    store[("store<br/>(Postgres, RLS per user)")]
+    engine["Summarization engine<br/>(unchanged core)"]
+    secrets["SecretStore<br/>(per-user token refs)"]
+
+    browser --> oidc
+    oidc --> gate
+    gate --> pages
+    pages --> store
+    connect --> secrets
+    connect --> store
+    account --> store
+    account --> secrets
+    summarize -->|background| engine
+    summarize -->|HTMX status poll| store
+    engine --> store
+```
+
+- **Dex OIDC session layer** (`internal/web/oidc`) — **authenticate-only** (ADR-012). It proves
+  *who*; authorization/isolation is the DB's job (RLS), not the session's.
+- **Registration gate** — a Dex subject with no `users` row is routed to `/register`, which
+  creates the `users` row + the `user_identities` mapping (migration 004). Returning subjects
+  pass straight through.
+- **Web-initiated YouTube connect** — `/oauth/youtube/connect` → `/oauth/youtube/callback`
+  persists a **per-user** refresh-token ref (`youtube/<userID>/refresh_token`) via `SecretStore`
+  and a `video_connections` row (ADR-006, migration 005). Distinct from the CLI `tapir auth`.
+- **Account management** — `/account` offers disconnect and **delete account**. Delete removes
+  only Tapir-side state (cascade across the user's tables + secret refs); the shared Dex identity
+  is left intact (ADR-013).
+- **Immediate summarization** — the web "Summarize" button (`POST /v/{id}/summarize`) fires the
+  engine in a **background goroutine** inside `serve`; the page HTMX-polls `/v/{id}/status`,
+  showing a Charmbracelet spinner while in-flight (and an honest "queued/waiting" state under
+  rate-limiting — ADR-014).
+- **Summarization mode** — `users.auto_summarize` (migration 006). Default is **true** for new
+  users (migration 011, ADR-018); all existing rows were back-filled via migration 012. Auto:
+  new videos **published within the recency window** (`TAPIR_AUTO_SUMMARIZE_WINDOW`, default ~7d,
+  ADR-020) are summarized automatically; older videos are discovered and listed but wait for an
+  explicit "Summarize". Manual: new videos appear unsummarized; the button sets
+  `videos.summarize_requested`, which the next `tapir run` processes and clears. A manual request
+  bypasses the recency bound. Both the click path and the batch `tapir run` drive the same
+  unchanged engine.
+- **List surface (ADR-020)** — the list reads `ListVideos` ordered summarized-first, then
+  `published_at DESC NULLS LAST`. The web layer collapses the noise so summaries are not buried:
+  un-summarized videos older than the recency window fold into one "Show N older videos"
+  disclosure, and caption-less videos collapse to a single count line. Copy surfaces scarcity
+  honestly (queue counts, gradual-fill note) — it never implies the feed is fuller than it is.
+
+The engine, ports, and sink adapters are **untouched** by all of the above — the web surface only
+reads the store and triggers the existing engine. Adding it changed wiring, not the core (ADR-003).
+
+---
+
+## In-process scheduler (ADR-018)
+
+`cmdServe` launches a background goroutine when `TAPIR_DISCOVERY_INTERVAL > 0`. On each tick
+it calls `store.ListAllUsers` (un-RLS'd admin query), builds a per-user `runner.Runner`, and
+calls `RunOnce` for each registered user in sequence.
+
+```mermaid
+sequenceDiagram
+    participant S as Scheduler goroutine
+    participant DB as Postgres (RLS)
+    participant YT as YouTube timedtext
+    participant LLM as LiteLLM gateway
+
+    loop every TAPIR_DISCOVERY_INTERVAL
+        S->>DB: ListAllUsers() [un-RLS'd]
+        loop per user
+            S->>DB: GetAutoSummarize(userID)
+            S->>YT: ListSubscriptions + NewVideos
+            Note over S,DB: auto: skip videos published before<br/>TAPIR_AUTO_SUMMARIZE_WINDOW (ADR-020);<br/>older ones listed, await manual request
+            Note over S,YT: WaitFetchGate(ctx) throttles<br/>all fetches to TAPIR_FETCH_RATE
+            alt transcript available
+                S->>LLM: Summarize
+                S->>DB: Deliver(summary)
+            else 429
+                S->>DB: SetTranscriptStatus(rate_limited)
+            end
+        end
+    end
+```
+
+**Single-replica constraint (load-bearing).** The scheduler lives in the web process;
+`replicas: 1` in the k3s deployment manifest is not cosmetic — running `tapir serve` at
+>1 replica makes every replica run the full discovery loop, causing every registered user
+to be fetched in parallel from the same egress IP (429s + duplicate work). Do not scale
+`serve` past 1 replica without first moving discovery to a k8s CronJob or adding leader
+election. The process logs a `Warn` at startup when scheduled discovery is enabled as a
+reminder.
+
+---
+
+## Process-wide timedtext rate gate
+
+**`internal/adapters/youtube/gate.go`** (ADR-014 item 2): a single `rate.Limiter`
+(`golang.org/x/time/rate`) shared across **all** Adapter instances. Every `httpDo` call for
+a caption fetch passes through `WaitFetchGate(ctx)` before hitting YouTube. This serialises
+the scheduler loop AND the web click-path through the same per-egress-IP budget. Configured
+via `TAPIR_FETCH_RATE` (Go duration, default `2s`). Setting it to `0` disables the gate
+(dev/tests only).
+
+This is the precondition that makes scheduled auto-summarize safe: without the gate, a
+multi-user scheduler pass could fire many concurrent timedtext requests from the same IP
+within seconds, triggering 429s for all users.
+
+### Two-path summarisation model
+
+Both paths share `globalFetchGate` — rate limiting is **respected in both**, not routed around.
+
+| Path | Trigger | Order | Rationale |
+|------|---------|-------|-----------|
+| **Foreground** | User clicks "Summarize" on any non-summarized card (`POST /v/{id}/retry-now` for rate-limited; `POST /v/{id}/summarize` for pending) | Single chosen video | On-demand value: user picks a specific video to read now — bypasses the recency bound |
+| **Background batch** | Scheduled discovery pass every `TAPIR_DISCOVERY_INTERVAL` | **Newest-first across all channels** (see below), **bounded to the recency window** (ADR-020) | Onboarding prioritisation within bounded load: recent videos auto-fill; the older back-catalogue stays on-demand |
+
+The rationale for both paths is **onboarding prioritisation under an honest, bounded load** — a
+new user gets summaries of their most recent videos automatically, while the older back-catalogue
+is listed but summarised only on demand, so it never re-drives the shared rate gate every cycle.
+
+### Newest-first batch ordering (ADR-018)
+
+Within each scheduled pass, `RunOnce` uses a three-phase structure:
+
+1. **Discover + persist**: walk all channels, `UpsertVideo` every candidate (so it appears in
+   the list), apply pre-filters (seen/manual/backoff/**recency**), collect surviving candidates.
+   The recency pre-filter (ADR-020) drops auto-mode videos published before
+   `now - TAPIR_AUTO_SUMMARIZE_WINDOW` unless they are explicitly requested; an undated video is
+   never aged out. They remain persisted/listed — only auto-summarisation is skipped.
+2. **Sort**: order candidates `published_at DESC, NULLS LAST, discovery_pos ASC`. Videos with
+   no publish date (schema 001: nullable) sort after all dated content. The sort is in-memory
+   (`slices.SortStableFunc`) — at current scale this is fine.
+3. **Process**: feed candidates to the engine in sorted order through `globalFetchGate`.
+
+Before (per-channel inline): `[chanA-old, chanA-mid, chanB-new, chanB-null]`
+After  (newest-first):       `[chanB-new, chanA-mid, chanA-old, chanB-null]`
+
+The set of *processed* videos now also excludes auto-mode back-catalogue beyond the recency
+window (those stay listed, summarised on demand); within the processed set, only order changes.
+
+---
+
 ## Sequence — core use case: new video summarized
 
 ```mermaid
@@ -191,6 +348,8 @@ Gherkin features in `docs/use-cases/`).
 
 - Audio-download + speech-to-text resolver (ADR-007) — would be an additional `VideoSource`
   fallback path, drawn when built.
-- Multi-tenant isolation primitives (per-tenant Postgres role, NetworkPolicy, tenant label)
+- Per-user isolation is **live, not deferred**: Postgres RLS `FORCE`d on every user-owned table
-  — activate at Stage 1 (ADR-002); single-user Stage 0 doesn't exercise them.
+  (ADR-012, migration 003), realising ADR-002's per-tenant intent at the DB layer. The coarser
+  multi-tenant primitives (per-namespace NetworkPolicy, Kyverno, tenant label) remain a
+  Stage-2 hardening item, not exercised yet.
 - Public SaaS surface (sign-up, billing) — Future C, not built (ADR-008).

docs/data-model.md

@@ -23,27 +23,39 @@ only opaque references to them; the secret material lives in ESO/1Password (ADR-
 
 ## Entities
 
+Solid entities below are **persisted today** (migrations 001–013). `AI_CREDENTIAL` and
+`SUBSCRIPTION` are **planned, not yet a table** — kept in the model for intent; see the notes.
+
 ```mermaid
 erDiagram
+    USER ||--|| USER_IDENTITY : "logs in via (Dex subject)"
     USER ||--o{ VIDEO_CONNECTION : has
-    USER ||--o{ AI_CREDENTIAL : has
+    USER ||--o{ SUMMARY_ACTION : records
-    VIDEO_CONNECTION ||--o{ SUBSCRIPTION : exposes
+    USER ||--o{ AI_CREDENTIAL : "has (planned)"
+    VIDEO_CONNECTION ||--o{ SUBSCRIPTION : "exposes (planned)"
     SUBSCRIPTION ||--o{ VIDEO : "produces (per user)"
     VIDEO ||--o| TRANSCRIPT : "has at most one"
     VIDEO ||--o| SUMMARY : "has at most one"
     SUMMARY ||--o{ SINK_DELIVERY : "delivered via"
+    USER ||--o{ CHANNEL_ERROR : "reports unavailable channels"
 
     USER {
         uuid id PK
         text display_name
+        bool auto_summarize "default true for new users (migration 011, ADR-018)"
+        timestamptz created_at
+    }
+    USER_IDENTITY {
+        text dex_subject PK
+        uuid user_id FK "UNIQUE -> USER, ON DELETE CASCADE; NOT RLS-enabled"
         timestamptz created_at
     }
     VIDEO_CONNECTION {
         uuid id PK
-        uuid user_id FK
+        uuid user_id FK "-> USER, ON DELETE CASCADE"
         text provider "youtube | vimeo"
-        text provider_account
+        text provider_account "nullable"
-        text token_secret_ref "-> SecretStore, never the token"
+        text token_ref "-> SecretStore, never the token"
         text status "active | revoked | error"
         timestamptz connected_at
     }
@@ -66,15 +78,18 @@ erDiagram
     }
     VIDEO {
         uuid id PK
-        uuid user_id FK
+        uuid user_id FK "-> USER, ON DELETE CASCADE"
-        uuid subscription_id FK
+        uuid subscription_id "nullable; no FK at Stage 0"
         text provider
         text provider_video_id
         text title
         int duration_s
         timestamptz published_at
         text url
+        bool summarize_requested "default false -> manual-mode queue flag (migration 006)"
         timestamptz seen_at
+        text transcript_status "none|rate_limited|fetched (migration 007)"
+        timestamptz rate_limited_at "backoff clock for 429 retries (migration 007)"
     }
     TRANSCRIPT {
         uuid video_id PK_FK
@@ -87,7 +102,7 @@ erDiagram
     SUMMARY {
         uuid id PK
         uuid user_id FK
-        uuid video_id FK
+        uuid video_id "no FK to videos; (user_id, video_id) UNIQUE is the dedup key"
         text summary
         jsonb highlights
         jsonb takeaways
@@ -98,26 +113,66 @@ erDiagram
     }
     SINK_DELIVERY {
         uuid id PK
-        uuid summary_id FK
+        uuid summary_id FK "-> SUMMARY, ON DELETE CASCADE; ownership derived via this FK"
         text sink "store | brain"
         text status "pending | delivered | error"
         text detail "nullable; error message etc"
         timestamptz updated_at
     }
+    SUMMARY_ACTION {
+        uuid id PK
+        uuid user_id FK "-> USER"
+        text video_id "TEXT, not FK (mirrors summaries' standalone key)"
+        text action "watched | skipped | saved"
+        timestamptz acted_at
+    }
+    CHANNEL_ERROR {
+        uuid user_id FK
+        text channel_id
+        text channel_name
+        timestamptz first_seen
+        timestamptz last_seen
+    }
 ```
 
+`SUMMARY_ACTION` has `UNIQUE (user_id, video_id, action)`; `VIDEO_CONNECTION` has
+`UNIQUE (user_id, provider)` (one connection per provider — reconnect upserts in place).
+RLS (`ENABLE` + `FORCE`) is on **every solid user-owned table above** — `users`, `videos`,
+`transcripts`, `summaries`, `summary_actions`, `video_connections`, `channel_errors`.
+`sink_deliveries` is RLS'd via an `EXISTS` on its parent summary; `user_identities` is
+intentionally **not** RLS'd (auth plumbing). See the *Isolation invariant* section for the
+mechanism.
+
 ## Notes per entity
 
-- **USER** — at Stage 0 there is exactly one row. At Stage 1, identity comes via Dex; this
+- **USER** — one row per registered user (Stage 1, ADR-012; no longer single-row). The Tapir-side
-  table holds the Tapir-side profile keyed to the Dex subject.
+  profile; the Dex identity is held separately in `USER_IDENTITY`, not on this row. `auto_summarize`
-- **VIDEO_CONNECTION** — a connected YouTube/Vimeo account. `token_secret_ref` resolves to
+  (migration 006) is the per-user mode flag: `TRUE` = auto-summarize new videos **published within
-  the OAuth refresh token via `SecretStore`. Revocation flips `status`, doesn't delete history.
+  the recency window** (`TAPIR_AUTO_SUMMARIZE_WINDOW`, default ~7d, ADR-020); older videos are
-- **AI_CREDENTIAL** — optional, per provider, per user (ADR-004's Fallback). Absent for users
+  listed but summarised on demand. Default is **true** for new users (migration 011, ADR-018);
-  who only use the local stack. One row per provider max.
+  existing rows were back-filled via migration 012 with RLS bypass.
-- **SUBSCRIPTION** — a watched channel. `websub_expires` tracks the YouTube push lease so the
+- **USER_IDENTITY** (migration 004) — the `dex_subject → user_id` map. `dex_subject` is the PK,
-  watcher knows when to re-subscribe; null for poll-based (Vimeo).
+  `user_id` a `UNIQUE` FK to `users` with `ON DELETE CASCADE`. This is the bridge resolved at login
+  *before* a `user_id` is known, so it is **deliberately not RLS-enabled** (it holds no user data;
+  RLS here would deadlock the lookup that yields the id used for scoping). Account deletion cascades
+  the mapping away (ADR-013).
+- **VIDEO_CONNECTION** (migration 005) — a connected YouTube/Vimeo account. `token_ref` resolves to
+  the OAuth refresh token via `SecretStore` (per-user scheme `youtube/<userID>/refresh_token`).
+  `UNIQUE (user_id, provider)`: one connection per provider, reconnect upserts. Revocation/disconnect
+  flips `status`, doesn't delete history. FORCE RLS'd.
+- **AI_CREDENTIAL** — *planned, no table yet.* Optional, per provider, per user (ADR-004's Fallback).
+  BYO keys are currently resolved via `SecretStore` refs without a dedicated table; this entity is
+  modelled for when per-credential metadata is needed.
+- **SUBSCRIPTION** — *planned, no table yet.* A watched channel; `websub_expires` would track the
+  YouTube push lease. At Stage 0/1 `videos.subscription_id` is a nullable column with **no FK** (the
+  subscriptions table is not part of the shipped store-sink slice — migration 001).
 - **VIDEO** — one row per (user, video) — note `user_id`, reflecting the per-user-isolation
   decision. The same video seen by two users is two rows. `seen_at` is when Tapir detected it.
+  `summarize_requested` (migration 006) is the manual-mode queue flag: the web "Summarize" button
+  sets it `TRUE`; the next `tapir run` picks it up, summarizes, and clears it back to `FALSE`.
+  `transcript_status` and `rate_limited_at` (migration 007) track caption-fetch outcomes for
+  rate-limit backoff: `NULL` = not attempted; `rate_limited` = 429 seen, skip until
+  `NOW() - rate_limited_at > TAPIR_FETCH_BACKOFF`; `fetched` = resolved; `none` = no transcript.
 - **TRANSCRIPT** — at most one per video. `source = none` records "checked, no usable
   transcript" so the watcher doesn't reprocess (ADR-007). `content` null in that case.
 - **SUMMARY** — at most one per video. `fallback_used` + `ai_provider`/`ai_model` make the
@@ -125,14 +180,46 @@ erDiagram
   `takeaways` as jsonb to stay schema-flexible while the output format settles.
 - **SINK_DELIVERY** — one row per (summary, sink) attempt. This is where "also sent to brain"
   lives — no brain tables, just a delivery row with `sink = brain`. Sinks fail independently;
-  a failed brain delivery doesn't fail the store delivery.
+  a failed brain delivery doesn't fail the store delivery. No own `user_id`; RLS ownership is
+  derived from the parent summary via `EXISTS` (migration 003).
+- **SUMMARY_ACTION** (migration 002) — records the maintainer's act on a summary (watch / skip /
+  save) — the column that makes the Stage-0 headline metric ("acts on ≥1 summary") queryable
+  (ui-spec.md §5, ADR-011). `video_id` is `TEXT` and **not** FK-constrained, mirroring summaries'
+  standalone `(user_id, video_id)` key. `UNIQUE (user_id, video_id, action)`. FORCE RLS'd.
+- **CHANNEL_ERRORS** (migration 013) — channels that returned HTTP 404 (deleted or private) on
+  the most recent discovery pass. Upserted per scheduler pass (`last_seen` refreshed each run);
+  surfaced on the account page as a warning. Cascades on user deletion. Primary key is
+  `(user_id, channel_id)`. FORCE RLS'd.
+- **LOGIN_EVENTS** (migration 010) — throttled one-row-per-(user, date) login stamp. Used by the
+  Stage-0 gate query (VISION §Stage 0: "returned and used in ≥2 distinct weeks").
 
-## Isolation invariant (Stage 1+)
+## Isolation invariant (Stage 1+) — LIVE
 
-Every user-owned table carries `user_id`. At Stage 1, this is enforced at the DB layer via a
+Every user-owned table carries `user_id`, and isolation is **enforced at the DB layer**, not
-per-tenant Postgres role + row grants (architecture review SC7), not only in application code.
+only in application code. ADR-011 shipped this surface single-user (one allowlisted subject,
-At Stage 0 (single user) the column exists but the enforcement is dormant. The isolation test
+enforcement dormant); **ADR-012 opened Stage 1 and turned enforcement on in the same slice.**
-in VISION Stage 2 asserts user A cannot read user B's rows.
+
+Enforcement is **Postgres Row-Level Security** (migration `003_rls.up.sql`):
+
+- RLS is `ENABLE`d **and** `FORCE`d on every user-owned table — `users`, `videos`,
+  `transcripts`, `summaries`, `summary_actions`, `video_connections`, `channel_errors`.
+  `FORCE` is load-bearing: the app connects as the table **owner** (`tapir` role), and owners
+  bypass RLS unless forced.
+- Each policy keys off the per-request GUC `tapir.current_user_id`, set transaction-locally by
+  the store's `withUser` helper via `set_config('tapir.current_user_id', $1, true)` — it
+  auto-resets on commit/rollback, so it never leaks across a pooled connection.
+- `current_setting('tapir.current_user_id', true)` uses `missing_ok = true`: an **unset** GUC
+  yields `NULL`, the predicate matches no rows, and access **denies by default**.
+- `sink_deliveries` has no `user_id`; its policy derives ownership from the parent summary via
+  `EXISTS (SELECT 1 FROM summaries …)`.
+- `user_identities` (the Dex-subject → user_id map) is **deliberately not RLS-enabled** — it is
+  auth plumbing read *before* a user_id is known; putting RLS there would deadlock. It holds no
+  user data.
+
+The Stage-2 isolation bar is **pulled forward, not deferred**: `internal/adapters/store/rls_test.go`
+runs two users against a non-superuser, non-`BYPASSRLS` role and asserts user A reads/writes zero
+of user B's rows across every table. It ships green with the multi-user features (ADR-012); no
+multi-user feature merges ahead of it passing.
 
 ## Job / processing state
 

docs/homelab-integration.md

@@ -2,7 +2,7 @@
 
 The concrete endpoints, conventions, and identifiers Tapir depends on, so an independent
 session doesn't have to rediscover them. **Verify anything marked "confirm" before relying on
-it** — endpoints and aliases drift, and this file is a snapshot (2026-06-02), not a live source.
+it** — endpoints and aliases drift, and this file is a snapshot (2026-06-06), not a live source.
 
 ## Local AI (the Primary in `llm.Router`)
 
@@ -160,5 +160,59 @@ allow per-provider when a user connects one.
 
 ---
 
-_Snapshot date 2026-06-02. Items marked **confirm** were not verified to a pinned source at
+_Snapshot date 2026-06-06. Items marked **confirm** were not verified to a pinned source at
 snapshot time — check brain or the live cluster before depending on them._
+
+## Stage 1 — multi-user facts (verified 2026-06-03)
+
+### Postgres RLS (ADR-012)
+- **The deployed DSN MUST connect as a non-superuser, non-BYPASSRLS role.** The
+  app uses the `tapir` role (table owner, non-superuser). `FORCE ROW LEVEL
+  SECURITY` is applied on all user-owned tables; a superuser DSN silently bypasses
+  FORCE and isolation is dead in prod. Verify: `SELECT rolsuper FROM pg_roles
+  WHERE rolname = 'tapir'` must return `f`.
+- Scoping is via `set_config('tapir.current_user_id', $userID, true)` (transaction-
+  local, auto-resets on commit — never leaks across a pooled connection).
+
+### Per-user YouTube token persistence
+- Stage-1 uses the **file-backed SecretStore** at `TAPIR_SECRETS_FILE=/data/secrets.json`
+  mounted from a **PVC** (`tapir-secrets`, 64Mi, RWO). Tokens survive pod restarts.
+  Upgrading to an ESO-backed per-user SecretStore is backlog (infra#86).
+- Per-user token ref scheme: `youtube/<userID>/refresh_token` (Worker C, ADR-006).
+  The Stage-0 single ref `youtube/refresh_token` is no longer used by `serve`; it
+  remains valid for the CLI `tapir run` (single-user, host-side).
+
+### Web YouTube connect
+- Redirect URI (registered in Google OAuth client, type Web): `https://tapir.d-ma.be/oauth/youtube/callback`.
+- Config env: `TAPIR_YT_CONNECT_REDIRECT_URL=https://tapir.d-ma.be/oauth/youtube/callback`.
+  `TAPIR_YT_CLIENT_ID` / `TAPIR_YT_CLIENT_SECRET` from the Web client (not the Desktop client used for the CLI).
+
+### Identity resolution
+- `user_identities(dex_subject → user_id)` table is **intentionally NOT RLS-enabled**
+  (it's auth plumbing, holds no user data; data isolation is on the user-owned tables).
+  All data access after subject resolution goes through `withUser`.
+
+## Scheduled discovery (ADR-018, verified 2026-06-05)
+
+`tapir serve` runs discovery for **all users** in-process on a timer (no CronJob). Three env
+knobs plus one load-bearing deployment constraint:
+
+- `TAPIR_DISCOVERY_INTERVAL` — Go duration, e.g. `2h`. The cadence the serve process runs a
+  discovery pass for every registered user (run-once-on-startup, then every interval).
+  **Unset or `0` = disabled** (dev/tests never auto-fetch).
+- `TAPIR_FETCH_RATE` — Go duration, default `2s`. The **process-wide per-egress-IP caption-fetch
+  rate gate** (ADR-014 item 2). Every caption fetch — scheduler runners *and* the web "Summarize"
+  click-path — serialises through this one limiter so the pod cannot collectively trip 429s. `0`
+  = unlimited (dev/tests). This is the precondition that makes auto-summarize-on-a-schedule safe;
+  do not raise it aggressively without watching for 429s.
+- `TAPIR_FETCH_BACKOFF=4h` — per-video rate-limit retry window; default `1h`. A video that
+  returns HTTP 429 on a caption fetch is skipped for this duration before being retried. The
+  scheduler checks `NOW() - rate_limited_at > TAPIR_FETCH_BACKOFF` before attempting to fetch
+  a video marked `transcript_status = rate_limited`. Longer values reduce 429 pressure at the
+  cost of slower recovery after a throttling episode.
+- **SINGLE-REPLICA WARNING (load-bearing).** The scheduler lives in the web process, so
+  `replicas: 1` in the deployment manifest is load-bearing: running `tapir serve` at >1 replica
+  makes **every** replica run the discovery loop → every user fetched in parallel from the same
+  egress IP (429s + duplicate work). Do **not** scale `serve` past 1 replica without first moving
+  discovery to a k8s CronJob or adding leader election. The process logs a `Warn` at startup when
+  scheduled discovery is enabled, as a reminder.

docs/specs/landing-page-and-doc-reconciliation.md

@@ -0,0 +1,153 @@
+# Spec — Landing page + documentation reconciliation
+
+**Date:** 2026-06-03
+**Status:** Ready to build
+**Scope:** Two parallel workstreams — (A) a public landing page; (B) reconciling the
+requirements / use-case / architecture / data-model docs against the deployed reality
+(v0.4.0). These are separate concerns; do not let one worker do both, or the audit gets
+done cursorily.
+
+All work: read `CLAUDE.md` + `DECISIONS.md` first. TBD — commit directly to `main`, one
+logical change per commit, conventional commits, `task check` green before every commit.
+After editing any `.templ`, run `templ generate` (the repo commits both `views.templ` and the
+generated `views_templ.go`).
+
+---
+
+## Workstream A — Public landing page
+
+### Goal
+A public landing page at `/welcome`, in the established bubbletea aesthetic, that lets a
+visitor sign in (one Dex flow) and, if already logged in, jump to their Tapir page or log out.
+New public transport surface only — no engine/core change (ADR-003).
+
+### Verified facts (read from `internal/web/oidc/oidc.go` @ main — do not re-guess)
+- Auth endpoints are exactly `/auth/login`, `/auth/callback`, `/auth/logout`.
+- `isPublicPath(p)` = `p == "/healthz" || strings.HasPrefix(p, "/auth/")` — the single
+  public-route chokepoint inside `DexAuth.Middleware`.
+- `DexAuth.CurrentUser(r) (web.User, bool)` reads the session cookie and does NOT redirect —
+  this is the "peek" the landing page uses to branch logged-in vs logged-out.
+- `handleCallback` redirects to `/` on success (correct — leave as-is).
+- `handleLogout` currently redirects to `loginPath` (`/auth/login`) — this is wrong for this
+  feature (see A3).
+- There is NO separate "sign up" against Dex/OIDC: one authorization flow. Registration is
+  Tapir's own `/register` step (ADR-012), reached after first login for an unknown subject.
+
+### Tasks
+**A1 — make `/welcome` public.** In `oidc.go`, extend `isPublicPath`:
+```go
+func isPublicPath(p string) bool {
+    return p == "/healthz" || p == "/welcome" || strings.HasPrefix(p, "/auth/")
+}
+```
+
+**A2 — unauthenticated bare-`/` → `/welcome`; deep links unchanged.** In `DexAuth.Middleware`,
+the unauthenticated branch currently always calls `redirectToLogin`. Change it so that when
+`r.URL.Path == "/"` an unauthenticated visitor is redirected to `/welcome`; for any other
+guarded path keep `redirectToLogin` (so a shared `/v/{id}` deep link still bounces through Dex
+and returns to the destination). Keep the `isPublicPath` check first (redirect-loop guard).
+
+**A3 — logout lands on `/welcome`, not login.** In `handleLogout`, change the final redirect
+from `loginPath` to `/welcome`. As written it sends the user to `/auth/login`, which
+immediately starts a fresh Dex login — visibly failing to log out. This intentionally breaks
+the existing logout test (oidc_test.go) which asserts redirect to `/auth/login`; update that
+test to expect `/welcome`. That break is expected, not a regression.
+
+**A4 — mount the landing handler** in `internal/web/handlers.go` `Router()`, on `root`,
+OUTSIDE `Auth.Middleware`, alongside `/healthz`:
+```go
+root.HandleFunc("GET /welcome", a.handleWelcome)
+```
+`handleWelcome` peeks `a.Auth.CurrentUser(r)` and renders `WelcomePage(user, ok)`. Not behind
+`Auth.Middleware` or `registrationGate`.
+
+**A5 — `WelcomePage` templ component** in `views.templ`. Reuse the existing shared
+layout/header partial and the established aesthetic (#7653FC purple rounded ╭─╮╰─╯ box, pink
+tapir mascot, #0EF9B6 mint accents) — match the existing pages, do not reinvent styling.
+- Logged out (`ok == false`): tapir mascot + tagline; one primary CTA **"Get Started"** →
+  `/auth/login`; honest sub-text: "New here? You'll set up your account right after signing in
+  — returning users go straight through." One button only (see verified facts: no separate
+  Dex sign-up; two buttons to the same URL would mislead).
+- Logged in (`ok == true`): "Go to my Tapir" → `/`; "Log Out" → `/auth/logout`. May greet via
+  `user.Email`.
+
+**A6 — tests** (extend `handlers_test.go` patterns). Note `StubAuth.CurrentUser` always returns
+true; for the logged-out case use a fake Auth returning `(web.User{}, false)`.
+- `GET /welcome`, no session → "Get Started" → `/auth/login`.
+- `GET /welcome`, with session → "Go to my Tapir" + "Log Out".
+- Unauthenticated `GET /` → 302 `/welcome`.
+- Unauthenticated `GET /v/{id}` → still 302 `/auth/login` (deep link preserved).
+- Authenticated `GET /` → still serves the list, unchanged.
+- oidc: `handleLogout` → 302 `/welcome` (update the existing test).
+
+**A out of scope:** no Dex config change, no new auth/session logic, no sign-up backend.
+
+---
+
+## Workstream B — Documentation reconciliation
+
+### Why
+The guardrail docs were written before Stage 1 and the web surface. Several now describe the
+opposite of the deployed reality (v0.4.0). Stale guardrail docs are worse than none — a future
+cold session (human or agent) trusts them. This workstream brings requirements, use cases,
+architecture, and data-model back in sync with `main`. Each fix is one commit; cite the ADR or
+migration that is the source of truth.
+
+### Known drift to fix (verified this session — not exhaustive; the worker confirms against code)
+**B1 — `internal/web/auth.go` comments.** The `User.Subject` doc and package doc still say
+"single-user allowlist (ADR-011)" / "Stage-0". Code is multi-user (ADR-012). Update the
+comments to describe the current multi-user reality; reference ADR-012.
+
+**B2 — `docs/data-model.md` isolation status.** It says isolation enforcement is "dormant at
+Stage 0". It is now LIVE: Postgres RLS, `FORCE`d on all user-owned tables, with a passing
+two-user isolation test (ADR-012, migration 003). Rewrite that section to describe enforced
+RLS as the current state; keep the history honest (was dormant at Stage 0, enforced from
+Stage 1).
+
+**B3 — `docs/data-model.md` schema completeness.** The doc predates migrations 002–006. Add
+the entities/columns that now exist: `summary_actions` (002), RLS (003), `user_identities`
+(004, dex_subject→user_id), `video_connections` (005), `users.auto_summarize` +
+`videos.summarize_requested` (006). The ER section should match the live schema. Cross-check
+against `internal/adapters/store/migrations/*.up.sql` — those are ground truth.
+
+**B4 — `docs/architecture/architecture.md`.** Predates the entire web surface. Update the C4
+container diagram and text to include: `tapir serve` (HTMX+Templ web reader/writer), the Dex
+OIDC session layer (`internal/web/oidc`), registration gate, web-initiated YouTube connect,
+account management, and the immediate-processing path (web "Summarize" button → background
+goroutine → status poll). The engine/ports/sinks core is unchanged (ADR-003) — show the web
+surface as a new transport over the same core, not a core change.
+
+**B5 — `docs/use-cases/*.feature`.** Add scenarios for the behaviours now live and unspecced:
+register (new subject → registration → user row; returning user straight through), connect
+YouTube (web OAuth), disconnect, delete-account (cascade + secret purge, Dex untouched —
+ADR-013), manual-vs-auto summarize mode + the Summarize button, and the landing page
+(logged-out CTA; logged-in shortcuts). Keep them as executable-style Gherkin consistent with
+the existing files.
+
+**B6 — `DECISIONS.md` ADR ordering (cosmetic).** ADR-010 sits before ADR-009/011 (append
+order). Reorder to numeric while you're in the file. Pure tidy, no content change.
+
+**B7 — requirements check.** If a requirements doc exists (e.g. `docs/ui-spec.md`, referenced
+by ADR-011), reconcile it with what shipped: note where the build deviated (e.g. the spinner /
+immediate processing / summarize mode were beyond the original spec) so the spec reflects
+reality or explicitly records the deviation. Do not silently rewrite history — record
+deviations as deviations.
+
+### B working method
+- Source of truth order: migrations + code > ADRs > prose docs. When a prose doc disagrees
+  with code, the code wins and the doc is corrected (unless the code is the bug — then flag it,
+  don't quietly doc around it).
+- One logical doc per commit. Cite the ADR/migration that justifies each change in the commit
+  body.
+- This is an audit, not a rewrite: preserve the docs' structure and the "rejected alternatives
+  / history" honesty. The goal is *current and trustworthy*, not *pretty*.
+
+---
+
+## Coordination
+A and B touch mostly different files (A: oidc.go, handlers.go, views.templ, tests; B: docs/* +
+auth.go comments). The one overlap is `auth.go` (B1 edits comments) vs A (reads it) — no
+conflict. Run A and B in parallel; commit independently to `main`.
+
+If anything in B reveals that code, not docs, is wrong (e.g. an isolation gap, a migration that
+doesn't match the data-model intent), STOP and surface it — that's a finding, not a doc edit.

docs/specs/newest-first-ordering.md

@@ -0,0 +1,77 @@
+# Spec — Newest-first batch ordering + honest "Try now" / prioritisation docs
+
+> **Extended by ADR-020 (2026-06-08).** This spec covers the *batch processing* order within a
+> pass. ADR-020 adds (a) a recency pre-filter — auto mode skips videos published before
+> `TAPIR_AUTO_SUMMARIZE_WINDOW`, listed but summarised on demand — and (b) the same
+> `published_at DESC NULLS LAST` ordering on the **list read** (`ListVideos`), which previously
+> sorted by `seen_at`. See `DECISIONS.md` ADR-020.
+
+**Repo:** tapir · **Size:** small · **Solo session.**
+
+**Why.** Product intent (maintainer, 2026-06-06): a new user should get summaries of their
+**newest** videos quickly, while the older back-catalogue fills in behind — all within the one
+shared rate gate. Today the foreground path ("Try now" button) lets a user hand-pick a video,
+but the **background batch processes in subscription/channel order, not newest-first** — so a new
+user with a large candidate set sees the batch summarise whatever channel is first in their
+subscription list, not their newest videos. This slice makes the batch agree with the intent, and
+fixes the docs to describe the real rationale (onboarding prioritisation), not the
+traffic-disguising framing a prior session wrote.
+
+Read `CLAUDE.md` + `DECISIONS.md` (ADR-014, ADR-018) first. TBD, conventional commits,
+`task check` green per commit, `templ generate` if views change.
+
+## 1. Newest-first batch ordering (the build)
+In `internal/runner/runner.go` `RunOnce`: today the loop processes each video inline while
+walking subscriptions channel-by-channel (`for sub → NewVideos → for v → process`). Change so
+that, within a pass, **candidates are processed newest-first across ALL channels**:
+- Collect the candidate videos across channels first (after dedup/seen/manual/rate-limit
+  filtering as today), then **sort by `published_at` descending before processing**, then process
+  in that order through the engine + shared `globalFetchGate`.
+- **`published_at` is nullable** (schema 001). Sort **NULLS LAST** — videos with no publish date
+  must not jump ahead of dated newest videos. Decide a stable tiebreak (e.g. `seen_at DESC`) for
+  equal/again-null dates.
+- Keep all existing behaviour: per-item failure isolation, the rate-limit backoff skip, manual
+  mode, channel-unavailable handling, stats. Ordering is the only change — not what gets
+  processed, just the order.
+- At 868 candidates a collect-then-sort in memory is fine; do **not** build a streaming/external
+  sort. Keep it simple.
+- The shared rate gate (`globalFetchGate`) is unchanged and still governs fetch pacing — ordering
+  does not bypass or weaken it.
+
+**Optional (only if cheap and clearly correct):** a soft cap so the *first* pass for a brand-new
+user summarises the newest N (e.g. 20) quickly and defers the long tail to subsequent passes — so
+onboarding value lands fast without waiting for the whole sorted set. If this adds real
+complexity, SKIP it and just do the newest-first ordering; the ordering alone delivers the intent.
+
+## 2. Tests
+- Given candidates across multiple channels with mixed `published_at` (incl. some NULL), assert
+  the processing order is newest-first, NULLS LAST, with the chosen tiebreak. Use the existing
+  fake VideoStore/Processor pattern in `runner_test.go`.
+- Assert ordering does not change *which* videos are processed vs. today (same set, new order).
+- Rate-gate / backoff / manual-mode behaviour unchanged (existing tests stay green).
+
+## 3. Docs — describe the REAL rationale (replace prior framing)
+The "Try now" button and the discovery batch together implement **onboarding prioritisation**:
+foreground (user-clicked "Try now") summarises a specific video on demand; background batch
+summarises newest-first; both honour the shared rate gate. **Update the docs to state this intent
+— and explicitly REMOVE/replace any framing that describes "Try now" as making traffic "look
+organic to YouTube" or evading rate limits.** That is not the rationale. The rationale is: *get
+the user a few summaries of their newest, most relevant videos fast; process the back-catalogue in
+the background; always within the honest shared rate limit.* Rate limiting is **respected**, not
+evaded.
+- `docs/ui-spec.md`: "Try now" = on-demand foreground summarisation of a chosen (typically newer)
+  video; rationale = fast onboarding value, not traffic shaping.
+- `docs/architecture/architecture.md`: document the two-path model — foreground on-demand vs.
+  background newest-first batch, both through `globalFetchGate` — and the newest-first ordering.
+- Any requirements/use-case doc mentioning discovery order: state newest-first.
+- If a brain note or `wiki` entry captured the "looks organic" rationale, correct it there too.
+
+## Boundaries
+- Do NOT increase fetch rate or weaken the rate gate. Account-safety constraint stands: the
+  caption endpoint is unofficial (ADR-010) and must be treated with honest backoff, never evasion.
+- Do NOT touch RLS, credentials, or the Dex surface.
+- Ordering change is within a pass only — no persisted priority queue, no new table.
+
+## Out of scope
+Per-user configurable ordering; priority weighting beyond newest-first; the soft-cap if it proves
+non-trivial.

docs/specs/scheduled-discovery.md

@@ -0,0 +1,92 @@
+# Spec — In-process scheduled discovery + auto-summarize + rate-gate finish
+
+> **Extended by ADR-020 (2026-06-08).** Auto-summarize is no longer "every unseen video": the
+> scheduler now skips videos published before `TAPIR_AUTO_SUMMARIZE_WINDOW` (default ~7d) unless
+> explicitly requested, so a back-catalogue does not re-drive the rate gate every cycle. See
+> `DECISIONS.md` ADR-020.
+
+**Repo:** tapir · **Size:** medium · **Solo session** (not a swarm).
+
+**Why this exists.** The Stage-0 gate ("me or a friend returns and reads/acts in ≥2 separate
+weeks") cannot be met because the system is not usable *unprompted*: discovery (`tapir run`) is
+host-side manual, so a newly onboarded user sees an empty list and never comes back. This slice
+makes Tapir watch on its own — the thing that makes the gate experiment actually runnable.
+
+Read `CLAUDE.md` + `DECISIONS.md` (esp. ADR-012, ADR-014, and the new ADR-018) first. TBD —
+commit directly to `main`, one logical change per commit, conventional commits, `task check`
+green before each commit. `templ generate` if any view changes.
+
+## Decisions already made (do not reopen)
+- **In-process scheduler**, NOT a k8s CronJob (maintainer's call: simpler deploy, acceptable
+  coupling at 3 users). The known cost — discovery shares the web process's lifetime and egress
+  — is accepted and recorded in ADR-018.
+- **Auto-summarize ON** for the maintainer + onboarded friends (zero-friction: the list fills
+  and summarizes itself).
+- **Gate clock resets** to when this ships (ADR-018) — until unprompted use is possible, the
+  prior window measured nothing.
+
+## 1. In-process scheduled discovery (core)
+- In `tapir serve` startup, launch a background goroutine that runs discovery for ALL users on
+  an interval: env `TAPIR_DISCOVERY_INTERVAL` (Go duration, e.g. `2h`). **Unset or 0 = disabled**
+  (so dev/tests never auto-fetch).
+- **Reuse the existing `runner.Runner` + `Loop`/`RunOnce`. Do NOT write a new scheduler.** The
+  per-user Runner already exists; the new work is **iterating users** and running one pass each
+  per tick. Enumerate users from the un-RLS'd `user_identities` (the same enumerate-then-act
+  pattern the login_events gate query established), then run each user's pass **inside that
+  user's RLS scope** (`withUser`).
+- **Stateless timing:** run-once-on-startup, then every interval — exactly the existing `Loop`
+  shape. Do NOT persist schedule state; a pod restart just restarts the cycle. Acceptable at this
+  scale. Do not build cron-in-Go.
+- **Graceful shutdown:** the goroutine respects `ctx` cancellation so a pod term doesn't wedge.
+- **Failure isolation in the loop:** one user's pass failing (or one channel/video) must not
+  abort the other users or crash `serve` — log and continue. (RunOnce already collects per-item
+  errors; preserve that at the per-user level too.)
+
+## 2. Auto-summarize default ON for Future-B users
+- New registrations default `auto_summarize = true` (so onboarded friends get zero-friction);
+  keep the account-page toggle so a user can switch to manual. One-off update existing user rows
+  to `true` as well (maintainer + any current users).
+- Consequence (intended): scheduled discovery both discovers AND summarizes new videos — which
+  is the point, and is why §3 is mandatory in the same slice.
+
+## 3. Finish/confirm the ADR-014 shared per-egress-IP rate gate (NOW load-bearing)
+- In-process scheduling + auto-summarize + multiple users = all caption fetches leave the **one
+  web pod's egress IP**, concurrently with any live "Summarize" button clicks. The timedtext
+  endpoint rate-limits per IP (ADR-010/014). Without a shared gate this self-inflicts 429s every
+  cycle.
+- **Confirm in code whether ADR-014 item 2 (a single PROCESS-WIDE rate gate) exists.**
+  Reconciliation flagged it as possibly built only as per-*video* backoff. If it is not a
+  process-wide gate, **build it now**: ONE shared limiter (token-bucket / min-interval) that
+  every timedtext/caption fetch passes through — scheduler loop AND click-path alike. Per-process,
+  not per-user, not per-video.
+- Keep the existing per-video 429 backoff (`rate_limited_at` + retry window) — complementary: the
+  gate prevents tripping 429; the backoff handles it if one still happens.
+- Honest UX (ADR-014 item 3) still applies: a fetch waiting on the gate shows "queued/waiting",
+  never a stuck spinner.
+
+## 4. Gate-clock reset — already recorded in ADR-018; verify VISION reflects it
+- ADR-018 (committed) resets the Stage-0 3–4 week window to start when this ships, and revises the
+  check-in date. VISION Stage 0 carries a pointer to it. The build doesn't re-decide this; just
+  ensure nothing in docs still implies the clock started earlier.
+
+## Tests
+- **Scheduler:** fake clock + fake Runner → N users each get one pass per tick; one user's failure
+  doesn't stop the others; `ctx` cancel stops the loop; interval=0 disables it entirely.
+- **Rate gate:** concurrent fetches (scheduler + simulated click) are serialized/limited through
+  the ONE gate — assert max-in-flight / min-interval honored regardless of caller.
+- **Auto-summarize default:** new registration → `auto_summarize = true`; account-page toggle
+  still flips it.
+
+## Out of scope / known constraints
+- No CronJob / k8s objects (in-process chosen).
+- **SINGLE-REPLICA ASSUMPTION (load-bearing).** In-process scheduling means if `tapir serve` ever
+  runs >1 replica, every replica runs the discovery loop → every user fetched in parallel (429s +
+  duplicate work). At 3 users this is single-replica, fine — but the build MUST note this
+  constraint in ADR-018 / deploy docs so a future scale-up doesn't silently double-run.
+- No persisted schedules, no multi-pod coordination.
+
+## Fallback if the session runs short
+Ship the scheduler with **auto-summarize OFF** (discovery only; manual Summarize button) until the
+process-wide rate gate (§3) is confirmed/built. NEVER ship auto-summarize-on-a-schedule without the
+gate — that combination self-inflicts 429s for every user every cycle. Auto-summarize ON is gated
+on §3 being done.

docs/specs/stage0-usage-measurement.md

@@ -0,0 +1,77 @@
+# Spec — Stage 0 usage measurement (login events)
+
+**Date:** 2026-06-03
+**Status:** Ready to build · **Repo:** tapir · **Size:** small (one migration + middleware + query)
+**Why:** The Stage 0 gate (VISION, ADR-016) is *return usage in ≥2 separate weeks*. `summary_actions`
+captures *acts* (watch/skip/save) but not *reads* — a friend who logs in weekly and reads summaries
+without clicking anything is invisible. For a **reading** product that is the most important signal.
+This adds the missing data so the gate is measurable as written. Solo session, not a swarm.
+
+Read `CLAUDE.md` + ADR-016 first. TBD, conventional commits, `task check` green before each commit.
+
+## Scope (resist sprawl — this is NOT analytics)
+
+A lightweight, append-only record of *when each user was active*, enough to answer
+"returned/read in ≥N distinct weeks". Not page-level events, not click tracking, not a funnel.
+
+### 1. Migration — `login_events` (append-only)
+```
+login_events (
+  id        UUID PK default gen_random_uuid(),
+  user_id   UUID NOT NULL,          -- per-user; RLS like every user-owned table
+  seen_at   TIMESTAMPTZ NOT NULL default NOW()
+)
+INDEX (user_id, seen_at)
+```
+- **RLS:** `FORCE ROW LEVEL SECURITY`, same policy/pattern as the other user-owned tables (the
+  `tapir.current_user_id` GUC via the `withUser` seam — match migration 003). A reporting query that
+  needs cross-user counts runs as the owner/maintainer outside the per-user scope, or via a dedicated
+  read — decide consistently with how existing admin-ish reads are done.
+- Append-only: no updates, no deletes except the user-delete cascade. **Add to the delete-account
+  cascade** (ADR-013) — `login_events` has no FK (mirrors `summary_actions`), so `DeleteUser` needs an
+  explicit delete for it, and the delete test must assert it's covered. *Do not forget this* — it's the
+  exact footgun the last delete work caught.
+
+### 2. Middleware — throttled stamp
+- In the authenticated request path (after `CurrentUserID` resolves, inside the registration-gated
+  app — NOT on `/welcome`/`/healthz`/`/auth`), record one `login_events` row **per user per day**
+  (throttle: skip if a row exists for this user with `seen_at` ≥ start-of-today). One insert per active
+  day, not per request — keeps the table small and the signal clean.
+- Throttle check must itself be RLS-scoped (`withUser`). Keep it cheap (indexed lookup).
+
+### 3. Query — the gate report
+Provide a query (and optionally a tiny `tapir report` CLI subcommand or an admin page — your call,
+CLI is fine) answering, per user:
+```sql
+-- distinct active weeks from reads (login_events) AND acts (summary_actions), unioned
+WITH weeks AS (
+  SELECT user_id, date_trunc('week', seen_at) AS wk FROM login_events
+  UNION
+  SELECT user_id, date_trunc('week', acted_at) FROM summary_actions
+)
+SELECT user_id, COUNT(DISTINCT wk) AS active_weeks
+FROM weeks GROUP BY user_id
+ORDER BY active_weeks DESC;
+```
+Gate passes when any user_id (maintainer or friend) reaches `active_weeks >= 2` within the window.
+
+## Honesty caveats to carry (from VISION/ADR-016)
+- **"Unprompted" is not measurable here.** login_events records *that* a user returned, not *why*. A
+  nudged return looks identical to an organic one. This build does not close that gap and must not
+  claim to — the VISION measurement note stands: count returns, read a nudged return as weaker signal.
+  (If prompt-tracking is ever wanted, that's a separate decision, not this build.)
+- **Data accrues from deploy onward.** The gate window's read-data starts when this ships — so ship
+  soon (maintainer's call) rather than batching with the infra tooling session.
+- **`date_trunc('week')` is ISO/timezone-sensitive** and noisy at low volume (N=3). Two visits days
+  apart can fall in the same or different weeks. Acceptable, but don't over-read a single-week-margin
+  pass/fail.
+
+## Out of scope
+Page/event analytics; prompt-vs-organic tracking; dashboards beyond the one gate query; anything
+touching the engine or sinks (this is web/store only — ADR-003 holds).
+
+## Tests
+- Migration up/down; RLS on `login_events` (extend the two-user isolation test to cover it).
+- Throttle: N requests same day → 1 row; next day → 2nd row.
+- `DeleteUser` removes the user's `login_events` and leaves others' intact (extend the delete test).
+- The gate query returns correct distinct-week counts across a seeded reads+acts fixture.

docs/specs/video-card-states.md

@@ -0,0 +1,77 @@
+# Spec — Unify video-card states: one "Summarize now" verb, honest no-captions state
+
+> **Superseded in part by ADR-020 (2026-06-08).** The five card states still hold, but the copy
+> changed: the nudge verb is now **"Summarize"** (not "Summarize now"), the rate-limited state
+> reads **"In queue"** (not "Fetching soon…"), and the queued state reads **"summarizing
+> shortly"** (not "waiting for the next run"). The list also now collapses older un-summarized
+> and caption-less videos. See `DECISIONS.md` ADR-020 and `views.templ` (`VideoCard`) for the
+> current copy; this doc is kept as the original design record.
+
+**Repo:** tapir · **Size:** small, **view-layer only** (`views.templ` + a little CSS in
+`view.go`; regenerate `views_templ.go`). No handler, store, or DB change. The two existing
+handlers (`/summarize`, `/retry-now`) stay exactly as they are — only what the card *shows*
+changes.
+
+**Why.** The video card today presents two different verbs — "Try now" (`.btn-retry`, on
+rate-limited videos) and "Summarize" (`.btn-secondary`, on pending videos) — for what the user
+experiences as one intent: *"summarize this video now."* The user doesn't know or care about the
+internal pipeline state (rate-limited vs. manual-queue); two differently-labelled, differently-
+styled buttons leak that state machine into the UI as a choice. Also: a **no-captions video
+(`TranscriptStatus == "none"`) currently falls into the `else` branch and wrongly shows a
+"Summarize" button** that, if clicked, tries and fails — there are no captions to fetch. That's
+the confusing dead-end to remove.
+
+In normal use the maintainer is in **auto mode**, so these buttons are *exceptions*, not the main
+path — videos summarize themselves. So the card should be **status-first**: the state is what the
+user reads constantly; the manual nudge is a small, quiet affordance for impatience, not a
+prominent call-to-action.
+
+## The honest per-state card model (footer of `VideoCard`)
+
+Restructure the footer branch in `VideoCard` (in `internal/web/views.templ`) to these states.
+The branch ORDER matters (summarized first, then terminal/no-action states, then actionable):
+
+1. **Summarized** — preview + provider chip + fallback badge + actions. **No button.** (unchanged)
+2. **No captions** (`r.TranscriptStatus == "none"`) — **NEW branch.** Quiet status text, e.g.
+   `No transcript available` (use a muted `.card-state`/`.chip-retry`-style treatment, NOT a
+   button). This is a terminal honest dead-end — the user can do nothing, so offer nothing.
+3. **Queued / requested** (`r.SummarizeRequested`) — "Queued · waiting for the next run". **No
+   button.** (unchanged)
+4. **Rate-limited** (`r.TranscriptStatus == "rate_limited"`) — quiet status (keep the
+   "fetching soon" sense) + a **quiet "Summarize now"** button POSTing to `retryNowURL` (clears
+   backoff then processes). Same quiet style as state 5.
+5. **Pending** (else — discovered, not yet attempted) — quiet "Not summarized" + a **quiet
+   "Summarize now"** button POSTing to `summarizeURL` (flips the queue flag then processes).
+
+## Unify the verb and the style
+- **One label everywhere a manual nudge is offered: "Summarize now"** (states 4 and 5). Drop the
+  "Try now" wording entirely.
+- **One quiet style** for both: use the understated `.btn-retry` pattern (small, pill, outline,
+  transparent bg) — NOT `.btn-secondary`/`.btn` (heavier). Rename the CSS class to something
+  state-neutral (e.g. `.btn-quiet` or `.btn-summarize-now`) so it no longer reads as
+  "retry"-specific; keep the same visual. The point: the nudge is subtle, status is primary.
+- Keep both `<form>`s posting to their respective existing handler URLs
+  (`retryNowURL` for rate-limited, `summarizeURL` for pending) with the existing HTMX
+  attributes (`hx-post`, `hx-target=#video-{id}`, `hx-swap=outerHTML`) — only the button
+  label/class change. The backend side-effect difference (clear-backoff vs. set-flag) stays
+  invisible to the user, which is correct.
+- Drop the engineer-facing `title="Fetch transcript now through the shared rate gate"` tooltip;
+  if a hint is wanted, make it user-facing ("Summarize this one now").
+
+## Quietness check (the design intent)
+The summary content and the per-state *status* are the card's primary information. The "Summarize
+now" button is a minor affordance. Do not make it a prominent solid-accent CTA — it must read as
+"you can nudge this if you're impatient", not "action required". Status text uses muted styling;
+the button uses the quiet outline style.
+
+## Tests
+- `videocard_internal_test.go` (exists): assert each of the 5 states renders the expected
+  footer — summarized (no button), no-captions (status, NO button, no `summarize`/`retry-now`
+  URL present), queued (no button), rate-limited ("Summarize now" → retry-now URL), pending
+  ("Summarize now" → summarize URL). The key new assertion: **a `none`-status video renders no
+  action button and no POST URL.**
+- Assert the label string "Try now" no longer appears anywhere in rendered output.
+
+## Out of scope
+Handler/DB changes; the detail-page action buttons (watched/skipped/saved — unrelated); the
+pipeline stats bar wording; auto/manual mode behaviour. Verb/label/style/no-captions-state only.

docs/ui-spec.md

@@ -72,7 +72,14 @@ summary_actions
   and join into the existing `SummaryRow` reads so list/detail show current state.
 - This column is what makes the Stage-0 metric ("did I act on a summary?") queryable.
 
-## 6. Auth (Dex OIDC, single-user authz)
+## 6. Auth (Dex OIDC)
+
+Authentication is delegated to the homelab OIDC provider at `TAPIR_OIDC_ISSUER` —
+**Authentik** since the Dex→Authentik migration (infra ADR-0001; ADR-019). It offers a
+Google upstream and Authentik-managed accounts (incl. its invite flow); Tapir no longer
+provisions accounts itself. Any authenticated subject can register a Tapir account
+(ADR-012: allowlist removed). The `oidc`/`DexAuth` package keeps its name for now (rename
+deferred, ADR-019).
 
 - **Flow:** standard Authorization Code. Use `coreos/go-oidc` + `golang.org/x/oauth2`
   (justify the deps in the commit; both are the homelab-standard OIDC libs and small).
@@ -92,7 +99,7 @@ summary_actions
 `TAPIR_OIDC_ISSUER` (`https://auth.d-ma.be`), `TAPIR_DEX_CLIENT_ID`, `TAPIR_DEX_CLIENT_SECRET`,
 `TAPIR_OIDC_REDIRECT_URL` (`https://tapir.d-ma.be/auth/callback`), `TAPIR_SESSION_SECRET`.
 Reuses existing `TAPIR_DB_DSN`, `TAPIR_USER_ID` (the StubAuth dev subject only). No secrets
-committed. (`TAPIR_ALLOWED_SUBJECT` was removed by ADR-012.)
+committed. (`TAPIR_ALLOWED_SUBJECT` was removed by ADR-012; use the keys above.)
 
 ## 8. Deployment — k3s + Flux GitOps
 
@@ -147,3 +154,29 @@ Gate (lane A) commits first; B/C/D follow.
 
 `task check` green per lane; B/C/D rebase on A. Deploy (D) lands last, after the binary serves
 locally.
+
+---
+
+## Deviations and additions (as-built)
+
+This spec describes the **Stage-0 single-user reader** (ADR-011). What actually shipped through
+v0.4.0 went further — Stage 1 (ADR-012) opened multi-user, and several UX features were added on
+top. Recorded here (append-only; the spec above is left intact) so intent and reality stay
+distinguishable.
+
+| As-built feature | What it is | Why | Covered by |
+|------------------|-----------|-----|------------|
+| **Multi-user + RLS isolation** | Several Dex users per deployment; isolation enforced by Postgres RLS, not the single-subject allowlist of §6. | Maintainer opened Stage 1 ahead of the formal Stage-0 gate, with DB-enforced isolation as the guardrail that keeps it safe. | ADR-012; migration 003 (`6775e5f`, `f28fdc0`, `2ae66da`) |
+| **Registration gate** | A Dex subject with no `users` row is routed to `/register`, which creates the `users` row + a `user_identities` mapping. (§2 listed "sign-up / user CRUD" as a non-goal.) | Explicit registration is how a multi-user surface stays honest — no just-in-time row creation. | ADR-012; `f396e01` |
+| **Per-user YouTube web connect** | `/oauth/youtube/connect` → `/oauth/youtube/callback` stores a per-user refresh-token ref + a `video_connections` row. (The spec assumed a host-side `tapir auth` only.) | Multi-user means each user connects their own account from the browser. | ADR-006, ADR-012; migration 005 (`0c9531a`, `2aad79b`) |
+| **Account management** | `/account` page with **disconnect** and **delete account**; delete removes only Tapir-side state and leaves the Dex identity intact. (§2 listed isolation/CRUD as non-goals.) | A real account needs a way out; deletion semantics are deliberately Tapir-side only. | ADR-013; `22eafcf`, `c7624d9`, `17d5e8c` |
+| **Immediate web summarization** | A quiet "Summarize now" button on non-summarized video cards. Pending cards POST to `/v/{id}/summarize` (queues + triggers engine); rate-limited cards POST to `/v/{id}/retry-now` (clears backoff + triggers engine). Both use the same `.btn-quiet` style and label — the internal pipeline distinction is invisible to the user. The page HTMX-polls `GET /v/{id}/status` while processing. Videos with `TranscriptStatus == "none"` show "No transcript available" with no button — this is a terminal honest state. (§2 said "triggering runs from the browser … do NOT build".) | Reading a list you can't act on is half a product; on-demand summarize closes the loop without waiting for a batch `tapir run`. | ADR-012, ADR-014; `25215cb`, `8c6c7ca` |
+| **Charmbracelet tapir spinner** | An animated in-flight indicator (charm palette) shown while a summarize is processing; an honest "queued/waiting" state under rate-limiting rather than a stuck spinner. | The spinner must tell the truth when the timedtext endpoint rate-limits (429), not imply imminence. | ADR-014; `25215cb`, `a4aeb5e` |
+| **Auto/manual summarization mode** | Per-user `auto_summarize`; manual lists new videos unsummarized and queues via `summarize_requested`; a mode toggle at `/account/summarize-mode`. Default is **true** for new users (migration 011, ADR-018); existing rows back-filled via migration 012. | Control over compute/noise — only summarize what the user cares about. | migration 006 (`748d5eb`, `bdbdce7`, `3014ee0`, `a269d4a`); migration 011/012 |
+| **Public landing page** | `/welcome` mounted **outside** the auth guard; unauthenticated `/` redirects there; logout returns there (not `/auth/login`). (The spec guarded everything except `/healthz` and `/auth/*`.) | A first-time visitor needs a public "what is this / get started" page before the login wall. | `d83943c`, `0fdf2f7`, `3a27bf1`, `d208110`, `8ca374e`, `f15f57f` |
+| **Invite onboarding** | **Removed from Tapir (ADR-019).** Invites are owned by the IdP (Authentik) now, not Tapir — the Dex local-password provisioning path (`tapir invite` CLI, `/invite/{token}` web flow, `internal/adapters/dex`) was deleted when the homelab migrated Dex→Authentik (infra ADR-0001). A new user is invited via Authentik's invite flow, logs into Tapir via OIDC, and is captured by the existing `/register` (display-name) gate. | Onboarding belongs to the identity provider; keeps Tapir out of the shared identity provider's write path. | ADR-019; infra ADR-0001 |
+| **Summarized-only filter** | `?summarized=1` query param on the list view. When set, only videos with a completed summary (`SummaryRow.Summarized = true`) are shown. Rendered as a "Summarized only" checkbox in the filter form. Summarized videos also sort to the top of the unfiltered list (`ORDER BY (s.id IS NOT NULL) DESC, seen_at DESC`). | Lets users focus on videos that are ready to read; newly landing summaries are visible at the top without filtering. | `internal/web/view.go` (`Filter.OnlySummarized`, `ListVideos` ORDER BY) |
+| **"Summarize now" foreground path** | Unified quiet nudge button on actionable non-summarized cards. Five explicit card states — (1) summarized: chip + no button; (2) no captions (`transcript_status = 'none'`): "No transcript available", no button; (3) queued: "Queued" chip, no button; (4) rate-limited: "Fetching soon…" + "Summarize now" → `POST /v/{id}/retry-now` (clears `rate_limited_at`, triggers engine); (5) pending: "Not summarized" + "Summarize now" → `POST /v/{id}/summarize` (queues + triggers engine). One verb, one style (`.btn-quiet`); backend difference invisible to user. Both handlers call `ProcessVideo` through `globalFetchGate`. Rate gate respected, not bypassed — this is onboarding prioritisation. | Fast onboarding value; honest dead-end for no-captions videos (no button that fails). | `internal/web/handlers.go` (`handleRetryNow`, `handleRequestSummarize`); `internal/web/views.templ` (`VideoCard`) |
+| **Pipeline stats bar** | A one-line status bar above the video list: `N summarized · M fetching soon · K no captions`. Computed from the unfiltered row set; hidden when all videos are summarized. Gives the user a clear read on pipeline state without any interaction. | Replaces the "why is nothing happening?" confusion when most videos are pending or rate-limited. | `internal/web/view.go` (`PipelineStats`, `pipelineStats`) |
+| **Unavailable channels (account page)** | The `/account` page shows a "Unavailable channels" section when any channels returned HTTP 404 on the last discovery pass. Lists channel name, an "unavailable" badge, and the first-seen date. Data sourced from the `channel_errors` table (migration 013). | Surfaces silent failures so users know why some subscribed channels produce no new videos. | migration 013; `internal/web/account.go`; `internal/adapters/youtube/youtube.go` (`domain.ErrChannelUnavailable`) |
+| **Recency window + sparse-state honesty (ADR-020)** | Supersedes the copy/sort in the rows above. Auto-summarize is bounded to videos published within `TAPIR_AUTO_SUMMARIZE_WINDOW` (~7d); older un-summarized videos collapse behind a single "Show N older videos — summarize on demand" disclosure, and caption-less videos collapse to a one-line count (not N cards). List order is now `summarized-first, published_at DESC NULLS LAST`. Copy reframed for honest scarcity: pipeline bar reads "N ready · M in queue · K no captions" (no "fetching soon"); a gradual-fill note explains the rate limit; the nudge verb is "Summarize" (not "Summarize now"); the queued card says "summarizing shortly"; the empty-connected state drops the impossible `tapir run` instruction. Detail leads with Takeaways. Filters slimmed (no date pickers; hidden when empty); watched/skipped segmented; back link on detail; empty terms checkbox removed. | Make the sparse reality legible and honest instead of implying abundance/imminence; bound auto load so the back-catalogue doesn't re-drive the caption gate. Never fetch harder — scarcity is surfaced, not engineered around. | ADR-020; `2384c47`, `3df0459`, `40b703e`, `a1a5217`, `4a0a56e`, `9bf1c31`, `980638d`, `12fb031`, `f775441`, `51aa5d9` |

docs/use-cases/connect_account.feature

@@ -10,12 +10,20 @@ Feature: Connect and manage video accounts
     And my refresh token is stored only as a secret reference
     And my subscriptions are synced
 
+  @pending
+  # Vimeo connect is not built yet (provider label exists; no connect flow or test).
   Scenario: Connect a Vimeo account
     Given I have no connected video accounts
     When I connect my Vimeo account
     Then the connection is stored with status "active"
     And my subscriptions are synced
 
+  Scenario: Connecting an account discovers videos immediately
+    Given I have no connected video accounts
+    When I connect my YouTube account
+    Then a discovery pass for my account is triggered right away
+    And I do not have to wait for the next scheduled pass to see my videos
+
   Scenario: Tokens are never stored in the clear
     When I connect any video account
     Then no OAuth token value is stored in the database
@@ -28,6 +36,9 @@ Feature: Connect and manage video accounts
     And no new videos are watched for that connection
     And my existing summaries remain readable
 
+  @pending
+  # Per-provider BYO credential config is not built as a web flow yet (the summarizer
+  # supports a fallback endpoint, but there is no user-facing BYO setup + its test).
   Scenario Outline: BYO AI credential is optional and per-provider
     When I configure a BYO provider "<provider>"
     Then the credential is stored only as a secret reference

docs/use-cases/landing_page.feature

@@ -0,0 +1,31 @@
+Feature: Public landing page
+  As a first-time visitor
+  I want a public welcome page before I log in
+  So that I understand what Tapir is and how to get started without hitting a login wall
+
+  Scenario: An unauthenticated visit to the root is sent to the welcome page
+    Given I am not logged in
+    When I open the root path "/"
+    Then I am redirected to "/welcome"
+
+  Scenario: The welcome page invites an unauthenticated visitor to start
+    Given I am not logged in
+    When I open "/welcome"
+    Then I see a "Get Started" call to action
+
+  Scenario: An authenticated user on the welcome page sees their way in and out
+    Given I am logged in
+    When I open "/welcome"
+    Then I see a link to my summaries
+    And I see a way to log out
+
+  @pending
+  # Behaviour ships (logout redirects to /welcome) but is not unit-tested: logout lives in
+  # the OIDC Auth impl and StubAuth has no routes to exercise it cheaply.
+  Scenario: Logging out returns to the welcome page
+    Given I am logged in
+    When I log out
+    Then I am returned to "/welcome"
+
+  # /welcome is mounted outside the auth guard so it is reachable without a session;
+  # the root and all data routes stay behind it (commits around the WelcomePage work).

docs/use-cases/registration.feature

@@ -0,0 +1,48 @@
+Feature: Register and manage a multi-user account
+  As one of a handful of trusted users
+  I want my own account, isolated from everyone else's
+  So that Tapir can serve several people from one deployment without leaking data
+
+  # Stage 1 (ADR-012): Dex authenticates, Tapir authorizes per user. A Dex subject
+  # with no users row is a new user and must register before reaching any data.
+
+  Scenario: A new Dex subject is routed to registration
+    Given I am authenticated by Dex with a subject that has no Tapir account
+    When I open any page that requires an account
+    Then I am routed to the registration page
+    And no summaries are shown until I register
+
+  Scenario: Registering creates the account and its identity mapping
+    Given I am authenticated by Dex with a subject that has no Tapir account
+    When I complete registration
+    Then a user row is created for me
+    And a user_identities row maps my Dex subject to that user
+    And I am taken into the app as a registered user
+
+  Scenario: A returning subject passes straight through
+    Given I am authenticated by Dex with a subject that already has a Tapir account
+    When I open the app
+    Then I am not asked to register again
+    And I see my own summaries
+
+  Scenario: Deleting an account removes only my data and leaves other users untouched
+    Given I am a registered user with summaries, a connected account, and recorded actions
+    And another user exists with their own summaries
+    When I delete my account
+    Then all of my rows are removed across every user-owned table
+    And my stored secret references are removed
+    And the other user's data remains intact
+    And my Dex identity is left intact
+
+  @pending
+  # Re-registration after delete is supported by design (delete leaves the Dex identity,
+  # ADR-013) but has no dedicated end-to-end test yet.
+  Scenario: A deleted user can register again as a fresh account
+    Given I deleted my Tapir account but my Dex identity still exists
+    When I sign in again
+    Then I am routed to the registration page as a new user
+    And registering creates a fresh user row with none of my old data
+
+  # Isolation is DB-enforced (Postgres RLS, ADR-012, migration 003): a user can never
+  # read or write another user's rows even if an application WHERE clause is wrong.
+  # Deletion is Tapir-side only — the shared Dex directory is never modified (ADR-013).

docs/use-cases/summarize_mode.feature

@@ -0,0 +1,47 @@
+Feature: Choose how new videos get summarized
+  As a user who wants control over compute and noise
+  I want to pick whether new videos are summarized automatically or on demand
+  So that I only spend summarization on the videos I actually care about
+
+  Background:
+    Given I am a registered user with a connected video account
+
+  Scenario: Auto mode summarizes recent new videos automatically
+    Given my summarization mode is "auto"
+    When a subscribed channel posts a new video with captions within the recency window
+    Then Tapir summarizes it without my asking
+    And the summary appears in my list
+
+  Scenario: Auto mode lists older videos without summarizing them
+    Given my summarization mode is "auto"
+    When discovery finds a video published before the recency window
+    Then the video appears in my list with no summary
+    And it is not summarized automatically
+    And I can still summarize it on demand with "Summarize"
+
+  Scenario: Automatic is the default for a new user
+    Given I have just registered
+    Then my summarization mode is "auto"
+
+  Scenario: Manual mode leaves new videos unsummarized
+    Given my summarization mode is "manual"
+    When a subscribed channel posts a new video with captions
+    Then the video appears in my list with no summary
+    And nothing is summarized until I request it
+
+  Scenario: Requesting a summary in manual mode queues it for the next run
+    Given my summarization mode is "manual"
+    And a new video is in my list with no summary
+    When I click "Summarize" on that video
+    Then the video is marked as requested
+    And the next run summarizes it
+    And the request flag is cleared after it is processed
+
+  # auto_summarize is a per-user setting and summarize_requested is a per-video queue
+  # flag (migration 006). The web button sets the flag; `tapir run` processes both the
+  # auto videos and the manually queued ones, then clears the flag.
+  #
+  # Recency bound (ADR-020): in auto mode the scheduler only summarizes videos published
+  # within TAPIR_AUTO_SUMMARIZE_WINDOW (default ~7d); older videos are discovered and
+  # listed but wait for an explicit "Summarize" — so a back-catalogue does not re-drive
+  # the per-IP caption gate (ADR-014) every cycle. A manual request bypasses the bound.

docs/ux-review/UX-REVIEW-stage0-recency.md

@@ -0,0 +1,207 @@
+# Tapir — Heuristic Review (Stage-0, sparse + recency-bounded)
+
+_Findings document, not a build spec. The maintainer filters; a spec follows separately.
+Reviewed against `VISION.md` (Stage-0 gate), `docs/ui-spec.md`, `internal/web/views.templ` +
+`view.go`, the prior `UX-REVIEW.md` pass, and the current screenshots. Written for the product
+**as it actually is**: ~283 discovered, ~15 summarized, ~256 in queue behind a respected per-IP
+caption rate limit, ~12 no-captions; single-user (maintainer) with friends pending; recency-
+bounded auto-summarize about to ship (auto = recent ~7d, older browsable + manual on demand)._
+
+## Reviewer stance
+
+The Stage-0 gate is **return usage**. So every finding is judged by one question: does this make
+the maintainer (or a friend) come back to an honestly-sparse feed? The visual layer is already
+decent — dark mode, cards, the constrained reader were all fixed in the prior pass. The open
+problems are **expectation-setting, honesty-of-scale, and the recency feed** — not pixels.
+
+**The single biggest risk:** a new user connects, sees `15 summarized · 256 fetching soon`,
+nothing visibly moves, and never returns. The whole gate dies at that moment. Most P0s below
+attack that one moment.
+
+## Tagging
+
+- **`[NOW]`** — improves the product as it is today (sparse, recency-bounded, single-user). Ships
+  in the upcoming bundle.
+- **`[LATER]`** — improves the product we hope it becomes (abundance, engagement, multiple users).
+  Valuable but premature until real usage validates the core loop.
+
+Severity: 🔴 breaks the core loop · 🟠 hurts it · 🟡 noticeable · 🔵 polish.
+
+---
+
+## P0 — fix before/with the recency ship
+
+### 1. 🔴 Empty-connected state tells the user to run a CLI command they can't run — `[NOW]`
+**Problem.** After connecting YouTube, the empty list says: *"Run `tapir run` to discover your
+subscriptions."* A friend on the web has no shell. And post-ADR-018 discovery is an in-process
+scheduled loop — so the instruction is wrong *even for the maintainer*. It is the first thing a
+newly-onboarded user sees, and it is an impossible, stale instruction.
+**Principle.** Match between system and the real world; help users recognize, not be blocked
+(Nielsen #1, #2, #9).
+**Proposal.** Replace with a passive, honest "we're working" state: *"Your account is connected.
+Tapir is finding your subscriptions and fetching captions — summaries appear here gradually. Check
+back later."* No command. No imperative the user can't satisfy.
+**Evidence.** `views.templ` `summaryList` → `.empty-connected`.
+
+### 2. 🔴 No expectation set for *gradual* fill — the return loop breaks at the cliff — `[NOW]`
+**Problem.** Captions are rate-limited by design; the backlog trickles over days. Nothing tells
+the user this. A first visit shows few/no summaries and no "come back" framing. The Stage-0 gate
+is literally about returns, and the product never asks for one or explains why patience is
+warranted.
+**Principle.** Visibility of system status (#1). And: the gate can't be cleared if the UX doesn't
+survive first contact.
+**Proposal.** One honest sentence near the pipeline bar / empty state: *"Tapir fetches captions
+slowly on purpose, to respect YouTube's limits. New summaries land gradually — usually best to
+check back tomorrow."* Turns confusing emptiness into intentional design. Highest-leverage change
+in the review.
+
+### 3. 🟠 "256 fetching soon" overstates imminence — a lie of scale — `[NOW]`
+**Problem.** The pipeline bar and rate-limited cards both say "fetching soon." For 256 items
+behind a per-IP throttle, "soon" is false — they trickle over days/weeks. Exactly the abundance-
+implying language the mandate forbids, inverted: it makes the *queue* look imminent.
+**Principle.** Honesty of scarcity (project mandate); #1.
+**Proposal.** Relabel by scale. Bar: `15 ready · 256 in queue · 12 no captions`. Card state:
+"In queue" / "Waiting its turn", not "Fetching soon…". Reserve "soon" for items actually next.
+**Evidence.** `views.templ` `pipelineBar`; card State 4.
+
+### 4. 🟠 Recency boundary is invisible in the feed — `[NOW, ships with recency]`
+**Problem.** Once auto-summarize is bounded to ~7d, a 6-month-old pending video and a 2-day-old
+pending video render identically ("Not summarized" + button). But only one is in the auto path;
+the other will *never* process unless clicked. The user can't tell "be patient, this is coming"
+from "this is yours to trigger or ignore."
+**Principle.** Visibility of system status; predictability (#1).
+**Proposal.** Bucket the list into two sections: **Recent** (auto, will fill itself) and
+**Older — browse / summarize on demand**. Card status language should encode *which side of the
+line it's on*, not the pipeline internals. Central IA decision of the recency change.
+
+### 5. 🟠 The un-summarized mass buries the ~15 readable summaries — `[NOW]`
+**Problem.** ~268 of 283 cards are not readable (pending / queued / no-captions). Summarized-first
+ordering helps, but the page is still 95% noise below the fold. "Attention is the scarce resource"
+is the product's own principle — and the default view violates it.
+**Principle.** Aesthetic/minimalist design; signal-to-noise (#8).
+**Proposal.** Default view = readable summaries + the Recent bucket. Collapse the older
+un-summarized mass behind *"Show 256 older un-summarized videos."* Collapse the 12 no-caption
+videos into a single line: *"12 videos have no captions"* (terminal, never readable — they don't
+deserve 12 full cards).
+
+---
+
+## P1 — high, near-term
+
+### 6. 🟠 "Summarize now" overpromises against the rate gate — `[NOW]`
+**Problem.** Button says "Summarize now" (title: "Summarize this one now"). Backend queues it
+behind the shared per-IP gate. Click 10 older videos and they all sit at "Fetching soon…". The
+verb sells immediacy the system can't honor.
+**Principle.** Honesty; match system/reality (#1).
+**Proposal.** Drop "now" → "Summarize". On click the card should honestly become "Queued" (it
+already can). Optionally show queue position once the queue is real. Don't engineer the gate
+harder — just stop the verb from lying.
+**Evidence.** `views.templ` card States 4 & 5; `handlers.go` `handleRetryNow`,
+`handleRequestSummarize`.
+
+### 7. 🟠 Welcome-page copy is stale and misleading post-ADR-019 — `[NOW]`
+**Problem.** Landing sub-copy: *"If you have an invite link, it will set up your account
+automatically."* Invites moved to Authentik (ADR-019); Tapir no longer handles invite links. The
+"Get Started" button goes straight to OIDC. The copy promises a flow that no longer exists.
+**Principle.** Match between system and reality (#2); honesty.
+**Proposal.** Rewrite: *"Tapir is invite-only right now. If you've been invited, sign in below."*
+Single CTA. Also set the gradual-fill expectation here (ties to #2) so it lands before the wall,
+not after.
+**Evidence.** `views.templ` `WelcomePage` → `.welcome-sub`.
+
+### 8. 🟡 "Queued · waiting for the next run" leaks system jargon — `[NOW]`
+**Problem.** "the next run" exposes the discovery-loop concept; a user doesn't know what a "run"
+is.
+**Principle.** Speak the user's language (#2).
+**Proposal.** "Queued — summarizing shortly." Hide the scheduler.
+**Evidence.** `views.templ` card State 3.
+
+### 9. 🟡 Filters render before there's anything to filter — `[NOW]`
+**Problem.** The fresh/empty state shows the full Channel/From/To/Filter bar *above* "Connect
+YouTube." Power tooling stacked on top of the one action that matters.
+**Principle.** Progressive disclosure; minimalist design (#8).
+**Proposal.** Hide the filter bar when there are 0 rows (and arguably below ~20). Show the connect
+CTA alone.
+**Evidence.** `fixes/03-empty-fresh.png`; `ListPage` renders `filterForm` unconditionally.
+
+### 10. 🟡 Date-range filters are dead weight at this scale — `[NOW]` demote / `[LATER]` rebuild
+**Problem.** From/To date pickers + a free-text exact-match Channel field are corpus-scale tools.
+With 15 summaries they're noise; channel-as-freetext is unguessable. (Confirm the prior review's
+#7 is resolved — that `Channel` shows a real channel name, not the `provider` string; seeded
+screenshots suggest it is, live data may differ.)
+**Principle.** Match tool to task; minimalist design.
+**Proposal.** `[NOW]`: reduce to the "Summarized only" toggle (+ maybe channel chips derived from
+present rows). Drop date pickers until the corpus justifies them. `[LATER]`: real channel facets +
+search when there's volume.
+
+---
+
+## P2 — reading experience & polish
+
+### 11. 🟡 Detail leads with Summary; the attention-saving payload (Takeaways) is last — `[NOW]`
+**Problem.** Product promise is "decide what's worth your time." The element that answers that —
+Takeaways / verdict — sits at the bottom. The user reads a full summary to reach the point.
+**Principle.** Lead with the user's actual job-to-be-done.
+**Proposal.** Reorder or add a one-line TL;DR/verdict at top. Takeaways → Highlights → Summary, or
+a "Worth watching?" lede. Data already exists; a reorder, not new machinery.
+**Evidence.** `views.templ` `DetailPage`.
+
+### 12. 🔵 No "back to Summaries" on detail — `[NOW]`
+**Problem.** Only the brand returns home, losing any filter context.
+**Proposal.** Explicit "← Summaries" link on the detail page.
+
+### 13. 🔵 Auto/manual toggle copy will be wrong after recency — `[NOW, with recency]`
+**Problem.** Account copy: *"Automatic summarizes every new video as it is discovered."* Becomes
+false once auto is bounded to ~7d.
+**Proposal.** *"Automatic summarizes new videos from the last ~7 days. Older videos stay
+browsable — summarize them on demand."*
+**Evidence.** `views.templ` `AccountPage` summarization section.
+
+### 14. 🔵 Register step asks acceptance of nonexistent terms — `[NOW]`
+**Problem.** "I accept the terms of use" — no terms linked. For a friends-only tool, ceremony
+accepting nothing.
+**Proposal.** Either link real terms or drop the checkbox at this stage.
+**Evidence.** `views.templ` `RegisterPage`.
+
+### 15. 🔵 watched/skipped mutual exclusivity unsignaled — `[NOW]` low
+**Problem.** Three independent-looking buttons; watched↔skipped are exclusive (carryover from
+prior review #14).
+**Proposal.** Segmented control for watched/skipped; keep Saved separate.
+
+---
+
+## `[LATER]` — premature until the core loop is validated
+
+_(abundance / engagement / multiple users)_
+
+- **Return-nudges (digest email / push).** 🔴 **Caution, not just defer.** A notification that
+  drives returns *contaminates the exact signal the gate measures* — VISION wants *unprompted*
+  returns and admits it can't distinguish prompted from organic. Building a nudge now poisons the
+  experiment. Defer until after the gate reads. `[LATER]`
+- **Full-text search across summaries** — needs volume to matter. `[LATER]`
+- **Channel facets / saved filters / sorting** — corpus-scale tooling (#10). `[LATER]`
+- **Read/unread + "new since last visit"** — genuinely helps returns, but only meaningful once
+  there is throughput to be "new." `[LATER]`
+- **Saved/queue view, collections** — engagement surface; no payoff at 15 items. `[LATER]`
+- **Richer card previews (top-takeaway as preview, 2–3 lines)** — triage aid that only pays off
+  with many cards to triage. `[LATER]`
+- **Backlog progress heartbeat ("N summarized this week", queue burn-down)** — rewards returning,
+  but needs real throughput to show motion; a static "last updated X ago" is the only `[NOW]`-worthy
+  slice. `[LATER]`
+- **Onboarding tour / multi-step welcome** — over-built for one user + a few friends. `[LATER]`
+
+---
+
+## What's already right (don't regress)
+
+Pipeline-bar concept, the honest "No transcript available" terminal state, the queued/waiting
+spinner instead of a fake-imminent one, the Unavailable-channels surface, the constrained-width
+reader, dark mode, the account danger-zone behind a disclosure. The honesty instincts are present
+— the P0s are about making that honesty *legible and correctly-scaled*, not adding it.
+
+## Two judgment calls to settle first
+
+1. **#4 (Recent/Older split) and #5 (collapse the older mass) are one decision viewed twice.**
+   Settle the recency-feed IA once and both fall out.
+2. **The return-nudge caution (`[LATER]` list, item 1) is the one to put in writing now** — before
+   someone "helpfully" ships an email digest to juice the gate and destroys the signal.

go.mod

@@ -10,7 +10,9 @@ require (
 	github.com/golang-migrate/migrate/v4 v4.19.1
 	github.com/jackc/pgx/v5 v5.9.2
 	github.com/stretchr/testify v1.11.1
+	golang.org/x/crypto v0.45.0
 	golang.org/x/oauth2 v0.36.0
+	golang.org/x/time v0.15.0
 )
 
 require (

go.sum

@@ -91,6 +91,8 @@ go.opentelemetry.io/otel/trace v1.37.0 h1:HLdcFNbRQBE2imdSEgm/kwqmQj1Or1l/7bW6mx
 go.opentelemetry.io/otel/trace v1.37.0/go.mod h1:TlgrlQ+PtQO5XFerSPUYG0JSgGyryXewPGyayAWSBS0=
 go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
 go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
+golang.org/x/crypto v0.45.0 h1:jMBrvKuj23MTlT0bQEOBcAE0mjg8mK9RXFhRH6nyF3Q=
+golang.org/x/crypto v0.45.0/go.mod h1:XTGrrkGJve7CYK7J8PEww4aY7gM3qMCElcJQ8n8JdX4=
 golang.org/x/oauth2 v0.36.0 h1:peZ/1z27fi9hUOFCAZaHyrpWG5lwe0RJEEEeH0ThlIs=
 golang.org/x/oauth2 v0.36.0/go.mod h1:YDBUJMTkDnJS+A4BP4eZBjCqtokkg1hODuPjwiGPO7Q=
 golang.org/x/sync v0.18.0 h1:kr88TuHDroi+UVf+0hZnirlk8o8T+4MrK6mr60WkH/I=
@@ -99,6 +101,8 @@ golang.org/x/sys v0.41.0 h1:Ivj+2Cp/ylzLiEU89QhWblYnOE9zerudt9Ftecq2C6k=
 golang.org/x/sys v0.41.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
 golang.org/x/text v0.31.0 h1:aC8ghyu4JhP8VojJ2lEHBnochRno1sgL6nEi9WGFGMM=
 golang.org/x/text v0.31.0/go.mod h1:tKRAlv61yKIjGGHX/4tP1LTbc13YSec1pxVEWXzfoeM=
+golang.org/x/time v0.15.0 h1:bbrp8t3bGUeFOx08pvsMYRTCVSMk89u4tKbNOZbp88U=
+golang.org/x/time v0.15.0/go.mod h1:Y4YMaQmXwGQZoFaVFk4YpCt4FLQMYKZe9oeV/f4MSno=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=

internal/adapters/store/account.go

@@ -13,10 +13,11 @@ import (
 // Deleting the users row cascades (ON DELETE CASCADE) to videos, transcripts,
 // summaries (→ sink_deliveries), video_connections, and the user_identities map
 // — referential-integrity cascades bypass RLS, so a user's child rows are removed
-// even though the deleting connection is scoped. summary_actions is the exception:
+// even though the deleting connection is scoped. summary_actions and login_events
-// it carries a user_id but has NO foreign key to users (migration 002), so the
+// are the exceptions: each carries a user_id but has NO foreign key to users
-// cascade does not reach it; it is deleted explicitly in the same scoped
+// (migrations 002 and 010), so the cascade does not reach them; they are deleted
-// transaction. Deleting an absent user is a no-op (idempotent).
+// explicitly in the same scoped transaction. Deleting an absent user is a no-op
+// (idempotent).
 //
 // This is tapir-side only (decision 2026-06-03): it removes all tapir data; the
 // Dex login identity is left untouched — a later login simply re-enters
@@ -28,6 +29,10 @@ func (s *Store) DeleteUser(ctx context.Context, userID string) error {
 			`DELETE FROM summary_actions WHERE user_id = $1`, userID); err != nil {
 			return fmt.Errorf("store: delete summary_actions: %w", err)
 		}
+		if _, err := tx.Exec(ctx,
+			`DELETE FROM login_events WHERE user_id = $1`, userID); err != nil {
+			return fmt.Errorf("store: delete login_events: %w", err)
+		}
 		if _, err := tx.Exec(ctx,
 			`DELETE FROM users WHERE id = $1`, userID); err != nil {
 			return fmt.Errorf("store: delete user: %w", err)

internal/adapters/store/channel_errors.go

@@ -0,0 +1,59 @@
+package store
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	"github.com/jackc/pgx/v5"
+)
+
+// ChannelError is a channel that returned HTTP 404 on a discovery pass.
+type ChannelError struct {
+	ChannelID   string
+	ChannelName string
+	FirstSeen   time.Time
+	LastSeen    time.Time
+}
+
+// UpsertChannelError records (or refreshes) a 404 channel for the current user.
+// Called by the runner inside a withUser scope; RLS guards user isolation.
+func (s *Store) UpsertChannelError(ctx context.Context, userID, channelID, channelName string) error {
+	return s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		_, err := tx.Exec(ctx, `
+			INSERT INTO channel_errors (user_id, channel_id, channel_name)
+			VALUES ($1, $2, $3)
+			ON CONFLICT (user_id, channel_id)
+			DO UPDATE SET channel_name = EXCLUDED.channel_name, last_seen = now()`,
+			userID, channelID, channelName)
+		if err != nil {
+			return fmt.Errorf("store: upsert channel error: %w", err)
+		}
+		return nil
+	})
+}
+
+// ListChannelErrors returns all 404-flagged channels for the user, newest first.
+func (s *Store) ListChannelErrors(ctx context.Context, userID string) ([]ChannelError, error) {
+	var out []ChannelError
+	err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		rows, err := tx.Query(ctx, `
+			SELECT channel_id, channel_name, first_seen, last_seen
+			FROM channel_errors
+			WHERE user_id = $1
+			ORDER BY last_seen DESC`, userID)
+		if err != nil {
+			return fmt.Errorf("store: list channel errors: %w", err)
+		}
+		defer rows.Close()
+		for rows.Next() {
+			var ce ChannelError
+			if err := rows.Scan(&ce.ChannelID, &ce.ChannelName, &ce.FirstSeen, &ce.LastSeen); err != nil {
+				return fmt.Errorf("store: scan channel error: %w", err)
+			}
+			out = append(out, ce)
+		}
+		return rows.Err()
+	})
+	return out, err
+}

internal/adapters/store/identity.go

@@ -31,6 +31,41 @@ func (s *Store) UserBySubject(ctx context.Context, subject string) (userID strin
 	return userID, true, nil
 }
 
+// UserIdentity is one (userID, dexSubject) pair from the un-RLS'd
+// user_identities map — the unit the scheduler enumerates to run a discovery
+// pass per user (ADR-018).
+type UserIdentity struct {
+	UserID     string
+	DexSubject string
+}
+
+// ListAllUsers returns every (userID, dexSubject) pair from user_identities. It
+// runs as a plain pool query WITHOUT withUser — intentional and legitimate:
+// user_identities is un-RLS'd auth plumbing (like UserBySubject), and the
+// scheduler enumerating all users to run their discovery passes is an admin
+// operation that cannot be scoped to any single user. Order is unspecified.
+func (s *Store) ListAllUsers(ctx context.Context) ([]UserIdentity, error) {
+	rows, err := s.pool.Query(ctx,
+		`SELECT user_id, dex_subject FROM user_identities`)
+	if err != nil {
+		return nil, fmt.Errorf("store: list all users: %w", err)
+	}
+	defer rows.Close()
+
+	var users []UserIdentity
+	for rows.Next() {
+		var u UserIdentity
+		if err := rows.Scan(&u.UserID, &u.DexSubject); err != nil {
+			return nil, fmt.Errorf("store: scan user identity: %w", err)
+		}
+		users = append(users, u)
+	}
+	if err := rows.Err(); err != nil {
+		return nil, fmt.Errorf("store: iterate user identities: %w", err)
+	}
+	return users, nil
+}
+
 // RegisterUser creates the tapir user for a Dex subject and the identity mapping
 // that points to it, returning the new user_id. It errors with
 // ErrSubjectRegistered if the subject already maps.

internal/adapters/store/identity_test.go

@@ -85,6 +85,39 @@ func TestRegisterUserRejectsDuplicateSubject(t *testing.T) {
 	require.Equal(t, first, got)
 }
 
+func TestListAllUsersReturnsEveryIdentity(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	empty, err := s.ListAllUsers(ctx)
+	require.NoError(t, err)
+	require.Empty(t, empty, "no registrations yet → empty slice")
+
+	const subjectC = "dex|carol-789"
+	idA, err := s.RegisterUser(ctx, subjectA, "Alice")
+	require.NoError(t, err)
+	idB, err := s.RegisterUser(ctx, subjectB, "Bob")
+	require.NoError(t, err)
+	idC, err := s.RegisterUser(ctx, subjectC, "Carol")
+	require.NoError(t, err)
+
+	users, err := s.ListAllUsers(ctx)
+	require.NoError(t, err)
+	require.Len(t, users, 3)
+
+	// Order is unspecified; compare as a set of (userID, subject) pairs.
+	got := make(map[string]string, len(users))
+	for _, u := range users {
+		got[u.DexSubject] = u.UserID
+	}
+	require.Equal(t, map[string]string{
+		subjectA: idA,
+		subjectB: idB,
+		subjectC: idC,
+	}, got)
+}
+
 func TestRegisterUserDistinctSubjectsGetDistinctUsers(t *testing.T) {
 	ctx := context.Background()
 	s := newStore(t)

internal/adapters/store/login.go

@@ -0,0 +1,41 @@
+package store
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/jackc/pgx/v5"
+)
+
+// StampLogin records that the user was active today, throttled to one row per
+// user per day. It is the read-side counterpart to SetAction: the middleware
+// calls it on every authenticated request, but the append happens at most once a
+// day so login_events stays small and the signal clean (one row = one active
+// day, not one request).
+//
+// The check-and-insert is a single atomic statement: the INSERT ... SELECT ...
+// WHERE NOT EXISTS only writes when no row for this user has seen_at in today
+// (date_trunc('day', NOW()), server timezone). It runs through withUser, so the
+// NOT EXISTS probe is itself RLS-scoped to the calling user via the
+// tapir.current_user_id GUC — one user's stamp can never be suppressed or
+// triggered by another user's rows. The explicit user_id predicate also keeps the
+// probe on the (user_id, seen_at) index.
+//
+// A unique constraint is deliberately not used: under concurrent same-day
+// requests the worst case is two rows for one day, which the gate query collapses
+// to a single week bucket anyway — not worth a write-blocking constraint.
+func (s *Store) StampLogin(ctx context.Context, userID string) error {
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		_, err := tx.Exec(ctx,
+			`INSERT INTO login_events (user_id)
+			 SELECT $1
+			 WHERE NOT EXISTS (
+			     SELECT 1 FROM login_events
+			      WHERE user_id = $1 AND seen_at >= date_trunc('day', NOW())
+			 )`, userID)
+		return err
+	}); err != nil {
+		return fmt.Errorf("store: stamp login: %w", err)
+	}
+	return nil
+}

internal/adapters/store/login_test.go

@@ -0,0 +1,83 @@
+package store_test
+
+import (
+	"context"
+	"testing"
+
+	"github.com/jackc/pgx/v5/pgxpool"
+	"github.com/stretchr/testify/require"
+)
+
+// countLoginEvents counts a user's login_events via the superuser pool, which
+// bypasses RLS — so the assertion sees the true row count regardless of scope.
+func countLoginEvents(t *testing.T, p *pgxpool.Pool, userID string) int {
+	t.Helper()
+	var n int
+	require.NoError(t, p.QueryRow(context.Background(),
+		`SELECT count(*) FROM login_events WHERE user_id = $1`, userID).Scan(&n))
+	return n
+}
+
+// TestStampLoginThrottlesToOnePerDay: repeated stamps within the same day insert
+// exactly one row — the throttle that keeps login_events one-row-per-active-day.
+func TestStampLoginThrottlesToOnePerDay(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	super := rawPool(t)
+	resetDB(t, super)
+	_, err := super.Exec(ctx, `INSERT INTO users (id) VALUES ($1)`, userA)
+	require.NoError(t, err)
+
+	for i := 0; i < 3; i++ {
+		require.NoError(t, s.StampLogin(ctx, userA))
+	}
+	require.Equal(t, 1, countLoginEvents(t, super, userA),
+		"three same-day stamps must collapse to one row")
+}
+
+// TestStampLoginRecordsOncePerNewDay: with yesterday's row already present, a
+// stamp today is NOT throttled — it appends the day's row, so distinct active days
+// accumulate (the substrate the gate's distinct-week count reads).
+func TestStampLoginRecordsOncePerNewDay(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	super := rawPool(t)
+	resetDB(t, super)
+	_, err := super.Exec(ctx, `INSERT INTO users (id) VALUES ($1)`, userA)
+	require.NoError(t, err)
+
+	// Seed an event dated yesterday (before today's start), so the throttle's
+	// "row exists with seen_at >= start-of-today" probe finds nothing for today.
+	_, err = super.Exec(ctx,
+		`INSERT INTO login_events (user_id, seen_at) VALUES ($1, NOW() - INTERVAL '1 day')`, userA)
+	require.NoError(t, err)
+
+	require.NoError(t, s.StampLogin(ctx, userA))
+	require.Equal(t, 2, countLoginEvents(t, super, userA),
+		"a stamp on a new day must append a second row")
+
+	// A second stamp the same day is throttled again.
+	require.NoError(t, s.StampLogin(ctx, userA))
+	require.Equal(t, 2, countLoginEvents(t, super, userA),
+		"the same-day repeat must not add a third row")
+}
+
+// TestStampLoginIsUserScoped: one user's stamp lands only on that user's rows —
+// the throttle probe is RLS-scoped, so user B's existing same-day row neither
+// suppresses nor is touched by user A's stamp.
+func TestStampLoginIsUserScoped(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	super := rawPool(t)
+	resetDB(t, super)
+	_, err := super.Exec(ctx, `INSERT INTO users (id) VALUES ($1), ($2)`, userA, userB)
+	require.NoError(t, err)
+
+	// B already has a same-day row; it must not throttle A's first stamp.
+	_, err = super.Exec(ctx, `INSERT INTO login_events (user_id) VALUES ($1)`, userB)
+	require.NoError(t, err)
+
+	require.NoError(t, s.StampLogin(ctx, userA))
+	require.Equal(t, 1, countLoginEvents(t, super, userA), "A's stamp must record despite B's same-day row")
+	require.Equal(t, 1, countLoginEvents(t, super, userB), "A's stamp must not touch B's rows")
+}

internal/adapters/store/migrate_test.go

@@ -0,0 +1,112 @@
+package store_test
+
+import (
+	"context"
+	"database/sql"
+	"os"
+	"testing"
+
+	"github.com/golang-migrate/migrate/v4"
+	migratepgx "github.com/golang-migrate/migrate/v4/database/pgx/v5"
+	"github.com/golang-migrate/migrate/v4/source/iofs"
+	"github.com/stretchr/testify/require"
+
+	_ "github.com/jackc/pgx/v5/stdlib" // register the "pgx" database/sql driver
+)
+
+// fileMigrator builds a golang-migrate instance from the on-disk migration files
+// (not the embedded FS the production Migrate uses), so a test can step the schema
+// up and down. os.DirFS(".") is rooted at the package dir; the SQL lives under
+// "migrations". Mirrors store.Migrate's construction otherwise.
+func fileMigrator(t *testing.T) *migrate.Migrate {
+	t.Helper()
+	db, err := sql.Open("pgx", dsn)
+	require.NoError(t, err)
+	t.Cleanup(func() { _ = db.Close() })
+
+	drv, err := migratepgx.WithInstance(db, &migratepgx.Config{})
+	require.NoError(t, err)
+	src, err := iofs.New(os.DirFS("."), "migrations")
+	require.NoError(t, err)
+	m, err := migrate.NewWithInstance("iofs", src, "pgx", drv)
+	require.NoError(t, err)
+	t.Cleanup(func() { _, _ = m.Close() })
+	return m
+}
+
+// loginEventsExists reports whether the login_events relation is present.
+func loginEventsExists(t *testing.T) bool {
+	t.Helper()
+	var reg *string
+	require.NoError(t, rawPool(t).QueryRow(context.Background(),
+		`SELECT to_regclass('public.login_events')::text`).Scan(&reg))
+	return reg != nil
+}
+
+// TestMigration010LoginEventsUpDown proves migration 010 is reversible: the down
+// migration drops login_events cleanly and the up migration recreates it. A rotten
+// down migration (forgotten DROP, dangling policy) would fail here rather than in
+// production during a rollback. The test restores the schema to latest before
+// returning so the shared embedded-postgres stays at HEAD for sibling tests.
+func TestMigration010LoginEventsUpDown(t *testing.T) {
+	newStore(t) // ensure the schema is migrated to latest (011 applied)
+	require.True(t, loginEventsExists(t), "login_events must exist at latest migration")
+
+	m := fileMigrator(t)
+	// 011, 012, 013 sit above 010; step them down first so 010 is exercised in isolation.
+	require.NoError(t, m.Steps(-1), "down 013 drops channel_errors, login_events intact")
+	require.True(t, loginEventsExists(t), "013 down leaves login_events intact")
+	require.NoError(t, m.Steps(-1), "down 012 is a no-op, login_events intact")
+	require.True(t, loginEventsExists(t), "012 down leaves login_events intact")
+	require.NoError(t, m.Steps(-1), "down 011 must not touch login_events")
+	require.True(t, loginEventsExists(t), "011 down leaves login_events intact")
+
+	require.NoError(t, m.Steps(-1), "down 010 must drop login_events")
+	require.False(t, loginEventsExists(t), "login_events must be gone after the down migration")
+
+	require.NoError(t, m.Steps(4), "up must recreate 010 then re-apply 011, 012, 013")
+	require.True(t, loginEventsExists(t), "login_events must be restored after the up migration")
+}
+
+// autoSummarizeDefault reads the users.auto_summarize column default as text
+// ("true"/"false"), so the migration's default flip is verifiable directly.
+func autoSummarizeDefault(t *testing.T) string {
+	t.Helper()
+	var def string
+	require.NoError(t, rawPool(t).QueryRow(context.Background(),
+		`SELECT column_default FROM information_schema.columns
+		  WHERE table_name = 'users' AND column_name = 'auto_summarize'`).Scan(&def))
+	return def
+}
+
+// TestMigration011AutoSummarizeDefaultUpDown proves migration 011 is reversible:
+// up sets the auto_summarize column default to TRUE (ADR-018), down restores
+// FALSE. The down intentionally does not revert existing rows — only the default.
+func TestMigration011AutoSummarizeDefaultUpDown(t *testing.T) {
+	newStore(t) // latest (013 applied)
+	require.Equal(t, "true", autoSummarizeDefault(t), "011 sets the default to TRUE")
+
+	m := fileMigrator(t)
+	require.NoError(t, m.Steps(-1), "down 013 drops channel_errors")
+	require.NoError(t, m.Steps(-1), "down 012 is a no-op")
+	require.NoError(t, m.Steps(-1), "down 011 reverts the column default")
+	require.Equal(t, "false", autoSummarizeDefault(t), "default is FALSE after the down migration")
+
+	require.NoError(t, m.Steps(1), "up 011 re-applies the TRUE default")
+	require.Equal(t, "true", autoSummarizeDefault(t))
+	require.NoError(t, m.Steps(1), "up 012 runs clean (no FORCE RLS on fresh schema)")
+	require.NoError(t, m.Steps(1), "up 013 creates channel_errors")
+}
+
+// TestMigration012FixAutoSummarizeRLS proves 012 runs cleanly and flips any
+// remaining auto_summarize=FALSE rows to TRUE (the back-fill blocked by RLS in 011).
+func TestMigration012FixAutoSummarizeRLS(t *testing.T) {
+	newStore(t) // apply all migrations including 012
+	require.Equal(t, "true", autoSummarizeDefault(t), "column default is TRUE after 012")
+
+	// Round-trip: down 012, then up 012 — must be idempotent.
+	m := fileMigrator(t)
+	require.NoError(t, m.Steps(-1), "down 012 must not error")
+	require.NoError(t, m.Steps(1), "up 012 must re-apply cleanly")
+	require.Equal(t, "true", autoSummarizeDefault(t), "default still TRUE after 012 re-applied")
+}

internal/adapters/store/migrations/006_summarization_mode.down.sql

@@ -0,0 +1,2 @@
+ALTER TABLE videos DROP COLUMN IF EXISTS summarize_requested;
+ALTER TABLE users DROP COLUMN IF EXISTS auto_summarize;

internal/adapters/store/migrations/006_summarization_mode.up.sql

@@ -0,0 +1,17 @@
+-- Migration 006: summarization mode (per-user auto/manual + per-video queue).
+--
+-- auto_summarize is a per-user setting (not a global one): multi-user ready per
+-- ADR-012. FALSE default makes MANUAL the out-of-the-box behavior — `tapir run`
+-- discovers new videos but only summarizes the ones the user explicitly queued.
+--
+-- summarize_requested is the per-video manual queue flag. The web "Summarize"
+-- button sets it TRUE; the next `tapir run` picks it up, summarizes, and clears
+-- it back to FALSE. In auto mode it is unused.
+--
+-- No RLS policy changes needed: both columns are added to tables that already
+-- carry user_id and have ENABLE + FORCE ROW LEVEL SECURITY (migration 003). A new
+-- column on an RLS-protected table inherits that protection automatically — the
+-- existing users_isolation / videos_isolation policies gate every row, so these
+-- columns are only ever readable/writable for the row's own user.
+ALTER TABLE users ADD COLUMN auto_summarize BOOLEAN NOT NULL DEFAULT FALSE;
+ALTER TABLE videos ADD COLUMN summarize_requested BOOLEAN NOT NULL DEFAULT FALSE;

internal/adapters/store/migrations/007_transcript_status.down.sql

@@ -0,0 +1,2 @@
+ALTER TABLE videos DROP COLUMN IF EXISTS rate_limited_at;
+ALTER TABLE videos DROP COLUMN IF EXISTS transcript_status;

internal/adapters/store/migrations/007_transcript_status.up.sql

@@ -0,0 +1,17 @@
+-- Migration 007: per-video transcript fetch status, for rate-limit backoff.
+--
+-- transcript_status records the outcome of the last transcript attempt:
+--   NULL          = not yet attempted
+--   'none'        = checked, no usable transcript (permanent — SourceNone)
+--   'fetched'     = transcript resolved and summarized (summary_id not null)
+--   'rate_limited'= the caption endpoint returned 429; retry after a backoff window
+--
+-- rate_limited_at stamps WHEN the 429 was seen, so the runner can skip re-fetching
+-- a still-throttled video until NOW() - rate_limited_at exceeds TAPIR_FETCH_BACKOFF.
+-- It is cleared (set NULL) whenever the status moves off 'rate_limited'.
+--
+-- No RLS policy changes needed: videos already has ENABLE + FORCE ROW LEVEL
+-- SECURITY (migration 003) with the videos_isolation policy. New columns inherit
+-- that protection automatically.
+ALTER TABLE videos ADD COLUMN transcript_status TEXT;
+ALTER TABLE videos ADD COLUMN rate_limited_at TIMESTAMPTZ;

internal/adapters/store/migrations/009_invitations.down.sql

@@ -0,0 +1 @@
+DROP TABLE IF EXISTS invitations;

internal/adapters/store/migrations/009_invitations.up.sql

@@ -0,0 +1,22 @@
+-- Migration 009: invitations — an email-based invite to join Tapir (Stage-1
+-- onboarding gate). Mathias mints one with `tapir invite <email>`; the recipient
+-- visits /invite/{token}, sets a password, and Tapir creates their Dex account.
+--
+-- Deliberately NOT user-owned and NOT under RLS: an invitation exists BEFORE the
+-- user does, so there is no user_id to scope by and no authenticated user context
+-- when the invite is created (host CLI) or consumed (public /invite handler, no
+-- Dex session). The token itself is the capability — a 32-byte crypto-random,
+-- single-use, time-boxed secret. Hence no `user_id` FK and no ENABLE/FORCE ROW
+-- LEVEL SECURITY here (unlike every user-owned table in migrations 003/005).
+CREATE TABLE invitations (
+    id         UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    email      TEXT NOT NULL,
+    token      TEXT NOT NULL UNIQUE,
+    created_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
+    expires_at TIMESTAMPTZ NOT NULL,
+    used_at    TIMESTAMPTZ
+);
+
+-- Lookups are by token (both the claim and the form preview); the UNIQUE
+-- constraint already creates an index, this names one explicitly for clarity.
+CREATE INDEX idx_invitations_token ON invitations(token);

internal/adapters/store/migrations/010_login_events.down.sql

@@ -0,0 +1,4 @@
+DROP POLICY IF EXISTS login_events_isolation ON login_events;
+ALTER TABLE login_events NO FORCE ROW LEVEL SECURITY;
+ALTER TABLE login_events DISABLE ROW LEVEL SECURITY;
+DROP TABLE IF EXISTS login_events;

internal/adapters/store/migrations/010_login_events.up.sql

@@ -0,0 +1,33 @@
+-- Migration 010: login_events records THAT a user was active (returned and read)
+-- on a given day — the Stage-0 signal summary_actions misses. summary_actions
+-- captures *acts* (watch/skip/save); a reader who logs in weekly and clicks
+-- nothing is otherwise invisible, yet for a reading product that return IS the
+-- signal the gate ("usage in >=2 distinct weeks", VISION/ADR-016) is defined on.
+--
+-- Append-only: one row per user per active day (the request-path throttle in the
+-- web layer enforces that cadence), never updated. Per-user isolation like every
+-- user-owned table.
+--
+-- NO foreign key to users (mirrors summary_actions, migration 002): user_id is
+-- carried for RLS/scoping but the table is decoupled so a stamp never blocks on a
+-- users row. The cost of that decoupling: the users-row cascade does NOT reach
+-- login_events, so DeleteUser must delete it explicitly (see account.go) — the
+-- exact footgun the summary_actions delete work caught.
+CREATE TABLE login_events (
+    id       UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    user_id  UUID NOT NULL,
+    seen_at  TIMESTAMPTZ NOT NULL DEFAULT NOW()
+);
+
+-- (user_id, seen_at) serves both the per-user-per-day throttle lookup
+-- (seen_at >= start-of-today) and the gate query's per-user week bucketing.
+CREATE INDEX idx_login_events_user_seen ON login_events(user_id, seen_at);
+
+-- RLS: identical GUC-keyed policy/pattern to migration 003. FORCE so the table
+-- owner (tapir, non-superuser in prod) is subject to it; an unset GUC yields NULL
+-- → no rows match → deny-all.
+ALTER TABLE login_events ENABLE ROW LEVEL SECURITY;
+ALTER TABLE login_events FORCE ROW LEVEL SECURITY;
+CREATE POLICY login_events_isolation ON login_events
+    FOR ALL
+    USING (user_id = current_setting('tapir.current_user_id', true)::uuid);

internal/adapters/store/migrations/011_auto_summarize_default.down.sql

@@ -0,0 +1,5 @@
+-- Revert the column default to FALSE. Existing rows are intentionally NOT
+-- reverted: flipping live users back to manual on a rollback would be a
+-- surprising regression (they may have come to rely on auto). The default change
+-- is the reversible part; data stays as the user left it.
+ALTER TABLE users ALTER COLUMN auto_summarize SET DEFAULT FALSE;

internal/adapters/store/migrations/011_auto_summarize_default.up.sql

@@ -0,0 +1,17 @@
+-- Migration 011: flip auto_summarize default to TRUE (ADR-018, Future-B).
+--
+-- Scheduled discovery (ADR-018) makes Tapir watch unprompted. For onboarded
+-- friends that only delivers zero-friction value if the list also SUMMARIZES
+-- itself — a manual default would mean the scheduler discovers videos a user
+-- still has to click through one by one, which is the empty-list problem again.
+-- So new users default to AUTO. The account-page toggle still lets a user switch
+-- to manual (SetAutoSummarize), so this only changes the out-of-the-box state.
+--
+-- Safe only because the process-wide caption-fetch rate gate (ADR-014 item 2)
+-- now exists: auto + scheduled + multi-user would otherwise self-inflict 429s
+-- every cycle. The gate is the precondition for shipping this default.
+ALTER TABLE users ALTER COLUMN auto_summarize SET DEFAULT TRUE;
+
+-- Bring existing rows (maintainer + any current registrations) onto the new
+-- default so they benefit immediately, not just users created after this point.
+UPDATE users SET auto_summarize = TRUE WHERE auto_summarize = FALSE;

internal/adapters/store/migrations/012_fix_auto_summarize_rls.down.sql

@@ -0,0 +1 @@
+-- No data revert: do not flip users back to manual on rollback.

internal/adapters/store/migrations/012_fix_auto_summarize_rls.up.sql

@@ -0,0 +1,6 @@
+-- Migration 011's UPDATE ran without tapir.current_user_id set, so FORCE RLS
+-- blocked all rows and zero users were updated. Temporarily drop FORCE so the
+-- table owner (tapir role) can bypass RLS for this back-fill, then restore it.
+ALTER TABLE users NO FORCE ROW LEVEL SECURITY;
+UPDATE users SET auto_summarize = TRUE WHERE auto_summarize = FALSE;
+ALTER TABLE users FORCE ROW LEVEL SECURITY;

internal/adapters/store/migrations/013_channel_errors.down.sql

@@ -0,0 +1 @@
+DROP TABLE IF EXISTS channel_errors;

internal/adapters/store/migrations/013_channel_errors.up.sql

@@ -0,0 +1,22 @@
+-- channel_errors: channels that returned HTTP 404 (deleted/private) on the most
+-- recent scheduler pass. Surfaced on the account page so users know why some
+-- subscribed channels produce no videos. Upserted per-pass; cleared when the
+-- channel starts returning results again (runner calls UpsertChannelError only
+-- on 404, so a recovered channel simply stops appearing after its row ages out
+-- or the user takes action). ON DELETE CASCADE keeps rows tidy on account deletion.
+CREATE TABLE channel_errors (
+    user_id      UUID        NOT NULL REFERENCES users(id)  ON DELETE CASCADE,
+    channel_id   TEXT        NOT NULL,
+    channel_name TEXT        NOT NULL,
+    first_seen   TIMESTAMPTZ NOT NULL DEFAULT now(),
+    last_seen    TIMESTAMPTZ NOT NULL DEFAULT now(),
+    PRIMARY KEY (user_id, channel_id)
+);
+
+CREATE INDEX idx_channel_errors_user_id ON channel_errors(user_id);
+
+ALTER TABLE channel_errors ENABLE ROW LEVEL SECURITY;
+ALTER TABLE channel_errors FORCE ROW LEVEL SECURITY;
+CREATE POLICY channel_errors_isolation ON channel_errors
+    FOR ALL
+    USING (user_id = current_setting('tapir.current_user_id', true)::uuid);

internal/adapters/store/reads.go

@@ -39,6 +39,20 @@ type SummaryRow struct {
 	FallbackUsed    bool
 	CreatedAt       time.Time
 	Actions         []string // current active actions for this video; nil when none
+
+	// Summarized reports whether a summary exists for this video. The summary-only
+	// reads (ListSummaries/GetSummaryByVideo) always yield true; the all-videos
+	// read (ListVideos) yields false for a discovered-but-unsummarized video, whose
+	// Summary/Highlights/AIProvider fields are then empty.
+	Summarized bool
+	// SummarizeRequested reflects videos.summarize_requested: the manual queue flag
+	// set by the web "Summarize" button and cleared by the next `tapir run`. Only
+	// populated by ListVideos/GetVideoRow (summary-only reads leave it false).
+	SummarizeRequested bool
+	// TranscriptStatus mirrors videos.transcript_status (migration 007): "" (unset),
+	// "none", "rate_limited", or "fetched". Drives the "Retrying later" list badge.
+	// Only populated by ListVideos/GetVideoRow ("" on summary-only reads).
+	TranscriptStatus string
 }
 
 // selectSummary is the shared projection for both reads. videos is LEFT JOINed
@@ -101,6 +115,161 @@ func (s *Store) ListSummaries(ctx context.Context, userID string, limit int) ([]
 	return out, nil
 }
 
+// selectVideo is the all-videos projection: it drives from the videos table and
+// LEFT JOINs the (at most one) summary, so a discovered-but-unsummarized video
+// still appears with empty summary fields. The column order mirrors selectSummary
+// for the shared fields, then appends summarized + summarize_requested. created_at
+// falls back to the video's seen_at when there is no summary, so the read-side row
+// always carries a sortable timestamp.
+const selectVideo = `
+	SELECT v.id,
+	       v.provider_video_id,
+	       COALESCE(v.title, ''),
+	       v.provider,
+	       COALESCE(v.url, ''),
+	       v.published_at,
+	       COALESCE(s.summary, ''),
+	       s.highlights,
+	       s.takeaways,
+	       COALESCE(s.ai_provider, ''),
+	       COALESCE(s.ai_model, ''),
+	       COALESCE(s.fallback_used, FALSE),
+	       COALESCE(s.created_at, v.seen_at),
+	       (s.id IS NOT NULL) AS summarized,
+	       v.summarize_requested,
+	       COALESCE(v.transcript_status, '')
+	  FROM videos v
+	  LEFT JOIN summaries s ON s.video_id = v.id AND s.user_id = v.user_id`
+
+// ListVideos returns ALL of the user's videos — summarized first, then by
+// published_at DESC with undated videos last, then seen_at DESC as a tiebreak —
+// capped at limit (non-positive defaults to 500). The published_at ordering
+// aligns the list with the recency framing (newest content first); seen_at
+// breaks ties and orders same/!undated rows deterministically. Unsummarized
+// videos come back with Summarized=false and empty summary fields, so the list
+// view can render them with a "Summarize" affordance. Scoped by user_id.
+func (s *Store) ListVideos(ctx context.Context, userID string, limit int) ([]SummaryRow, error) {
+	if limit <= 0 {
+		limit = 500
+	}
+	var out []SummaryRow
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		rows, err := tx.Query(ctx,
+			selectVideo+`
+			 WHERE v.user_id = $1
+			 ORDER BY (s.id IS NOT NULL) DESC, v.published_at DESC NULLS LAST, v.seen_at DESC
+			 LIMIT $2`,
+			userID, limit)
+		if err != nil {
+			return fmt.Errorf("store: list videos: %w", err)
+		}
+		defer rows.Close()
+
+		for rows.Next() {
+			row, err := scanVideoRow(rows)
+			if err != nil {
+				return err
+			}
+			out = append(out, row)
+		}
+		if err := rows.Err(); err != nil {
+			return fmt.Errorf("store: iterate videos: %w", err)
+		}
+		return nil
+	}); err != nil {
+		return nil, err
+	}
+	if err := s.attachActions(ctx, userID, out); err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// GetVideoRow returns a single video row (summarized or not) for (userID,
+// videoID), used to re-render one card after queuing it. Returns ErrNotFound when
+// the user has no such video. Scoped by user_id.
+func (s *Store) GetVideoRow(ctx context.Context, userID, videoID string) (*SummaryRow, error) {
+	var (
+		row   SummaryRow
+		found bool
+	)
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		rows, err := tx.Query(ctx,
+			selectVideo+`
+			 WHERE v.user_id = $1 AND v.id = $2`,
+			userID, videoID)
+		if err != nil {
+			return fmt.Errorf("store: get video: %w", err)
+		}
+		defer rows.Close()
+
+		if !rows.Next() {
+			if err := rows.Err(); err != nil {
+				return fmt.Errorf("store: get video: %w", err)
+			}
+			return nil
+		}
+		row, err = scanVideoRow(rows)
+		if err != nil {
+			return err
+		}
+		found = true
+		return nil
+	}); err != nil {
+		return nil, err
+	}
+	if !found {
+		return nil, ErrNotFound
+	}
+	holder := []SummaryRow{row}
+	if err := s.attachActions(ctx, userID, holder); err != nil {
+		return nil, err
+	}
+	return &holder[0], nil
+}
+
+// scanVideoRow reads one row in the selectVideo column order. published_at is
+// nullable so it scans through a pointer.
+func scanVideoRow(rows pgx.Row) (SummaryRow, error) {
+	var (
+		row         SummaryRow
+		highlights  []byte
+		takeaways   []byte
+		publishedAt *time.Time
+	)
+	if err := rows.Scan(
+		&row.VideoID,
+		&row.ProviderVideoID,
+		&row.Title,
+		&row.Channel,
+		&row.URL,
+		&publishedAt,
+		&row.Summary,
+		&highlights,
+		&takeaways,
+		&row.AIProvider,
+		&row.AIModel,
+		&row.FallbackUsed,
+		&row.CreatedAt,
+		&row.Summarized,
+		&row.SummarizeRequested,
+		&row.TranscriptStatus,
+	); err != nil {
+		return SummaryRow{}, fmt.Errorf("store: scan video: %w", err)
+	}
+	if publishedAt != nil {
+		row.PublishedAt = *publishedAt
+	}
+	var err error
+	if row.Highlights, err = unmarshalList(highlights); err != nil {
+		return SummaryRow{}, fmt.Errorf("store: unmarshal highlights: %w", err)
+	}
+	if row.Takeaways, err = unmarshalList(takeaways); err != nil {
+		return SummaryRow{}, fmt.Errorf("store: unmarshal takeaways: %w", err)
+	}
+	return row, nil
+}
+
 // GetSummaryByVideo returns the full summary for (userID, videoID), including
 // highlights and takeaways. Returns ErrNotFound when the user has no such
 // summary. Scoped by user_id.

internal/adapters/store/report.go

@@ -0,0 +1,111 @@
+package store
+
+import (
+	"context"
+	"fmt"
+	"sort"
+
+	"github.com/jackc/pgx/v5"
+)
+
+// UserActiveWeeks is one row of the Stage-0 gate report: how many DISTINCT
+// calendar weeks a user was active in, counting reads (login_events) AND acts
+// (summary_actions) together. The gate (VISION/ADR-016) passes when any user
+// reaches ActiveWeeks >= 2.
+type UserActiveWeeks struct {
+	UserID      string
+	DisplayName string
+	ActiveWeeks int
+}
+
+// ActiveWeeks computes per-user distinct-active-weeks for the gate report, most
+// active first.
+//
+// Why per-user iteration rather than one cross-user GROUP BY: the user-owned
+// tables are FORCE RLS (migration 003/010) and the production role is a non-
+// superuser owner, so a single un-scoped query sees nothing (deny-all). Instead we
+// enumerate users from the deliberately un-RLS'd identity map (user_identities,
+// migration 004) and count each user's weeks inside withUser, where the GUC scopes
+// login_events + summary_actions to that user. No privilege escalation, no policy
+// change — the same isolation seam every other read flows through.
+//
+// Scope note: the enumeration covers users with a Dex identity (the web users the
+// gate is about). A CLI-only user created by the store sink without an identity
+// row would not appear — out of scope for this gate.
+func (s *Store) ActiveWeeks(ctx context.Context) ([]UserActiveWeeks, error) {
+	userIDs, err := s.identityUserIDs(ctx)
+	if err != nil {
+		return nil, err
+	}
+
+	out := make([]UserActiveWeeks, 0, len(userIDs))
+	for _, uid := range userIDs {
+		row, err := s.activeWeeksFor(ctx, uid)
+		if err != nil {
+			return nil, err
+		}
+		out = append(out, row)
+	}
+
+	// Most active first; user_id as a stable tie-break for deterministic output.
+	sort.SliceStable(out, func(i, j int) bool {
+		if out[i].ActiveWeeks != out[j].ActiveWeeks {
+			return out[i].ActiveWeeks > out[j].ActiveWeeks
+		}
+		return out[i].UserID < out[j].UserID
+	})
+	return out, nil
+}
+
+// identityUserIDs lists every tapir user_id from the un-RLS'd identity map. It
+// runs directly on the pool (no withUser): user_identities carries no user data
+// and is intentionally not RLS-enabled, so it is the one table that can be read
+// pre-scope to discover who exists.
+func (s *Store) identityUserIDs(ctx context.Context) ([]string, error) {
+	rows, err := s.pool.Query(ctx, `SELECT user_id FROM user_identities ORDER BY user_id`)
+	if err != nil {
+		return nil, fmt.Errorf("store: list identity users: %w", err)
+	}
+	defer rows.Close()
+
+	var ids []string
+	for rows.Next() {
+		var id string
+		if err := rows.Scan(&id); err != nil {
+			return nil, fmt.Errorf("store: scan identity user: %w", err)
+		}
+		ids = append(ids, id)
+	}
+	if err := rows.Err(); err != nil {
+		return nil, fmt.Errorf("store: iterate identity users: %w", err)
+	}
+	return ids, nil
+}
+
+// activeWeeksFor counts one user's distinct active weeks (reads UNION acts) and
+// reads their display name, RLS-scoped via withUser. The UNION dedups a week that
+// has both a login and an action so it counts once.
+func (s *Store) activeWeeksFor(ctx context.Context, userID string) (UserActiveWeeks, error) {
+	res := UserActiveWeeks{UserID: userID}
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		if err := tx.QueryRow(ctx,
+			`WITH weeks AS (
+			     SELECT date_trunc('week', seen_at) AS wk
+			       FROM login_events WHERE user_id = $1
+			     UNION
+			     SELECT date_trunc('week', acted_at)
+			       FROM summary_actions WHERE user_id = $1
+			 )
+			 SELECT count(DISTINCT wk) FROM weeks`, userID).Scan(&res.ActiveWeeks); err != nil {
+			return fmt.Errorf("store: count active weeks: %w", err)
+		}
+		if err := tx.QueryRow(ctx,
+			`SELECT COALESCE(display_name, '') FROM users WHERE id = $1`, userID).Scan(&res.DisplayName); err != nil {
+			return fmt.Errorf("store: read display name: %w", err)
+		}
+		return nil
+	}); err != nil {
+		return UserActiveWeeks{}, err
+	}
+	return res, nil
+}

internal/adapters/store/report_test.go

@@ -0,0 +1,78 @@
+package store_test
+
+import (
+	"context"
+	"testing"
+
+	"github.com/jackc/pgx/v5/pgxpool"
+	"github.com/stretchr/testify/require"
+)
+
+// seedReportUser inserts a user + its identity mapping (the enumeration source
+// ActiveWeeks reads). display_name is optional.
+func seedReportUser(t *testing.T, p *pgxpool.Pool, userID, subject, name string) {
+	t.Helper()
+	ctx := context.Background()
+	_, err := p.Exec(ctx,
+		`INSERT INTO users (id, display_name) VALUES ($1, NULLIF($2, ''))`, userID, name)
+	require.NoError(t, err)
+	_, err = p.Exec(ctx,
+		`INSERT INTO user_identities (dex_subject, user_id) VALUES ($1, $2)`, subject, userID)
+	require.NoError(t, err)
+}
+
+// TestActiveWeeksCountsDistinctWeeksAcrossReadsAndActs is the gate-query proof.
+// It seeds, with fixed timestamps in known ISO weeks:
+//   - user A: reads in week of Jan 5 and Jan 12, acts in week of Jan 12 (dup) and
+//     Jan 19 → the UNION across both tables collapses the shared week → 3 distinct.
+//   - user B: a single read in the week of Jan 5 → 1 distinct (below the gate).
+//
+// It verifies the count is correct, dedups the cross-table shared week, and orders
+// most-active first.
+func TestActiveWeeksCountsDistinctWeeksAcrossReadsAndActs(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	p := rawPool(t)
+	resetDB(t, p) // TRUNCATE ... users CASCADE also clears user_identities
+
+	seedReportUser(t, p, userA, "subject-a", "Ada")
+	seedReportUser(t, p, userB, "subject-b", "")
+
+	// Reads (login_events) — fixed dates in distinct ISO weeks.
+	_, err := p.Exec(ctx,
+		`INSERT INTO login_events (user_id, seen_at) VALUES
+		   ($1, '2026-01-05T09:00:00Z'),
+		   ($1, '2026-01-12T09:00:00Z'),
+		   ($2, '2026-01-05T09:00:00Z')`, userA, userB)
+	require.NoError(t, err)
+
+	// Acts (summary_actions) — one in A's week-of-Jan-12 (shared with a read, must
+	// dedup) and one in a new week (Jan 19).
+	_, err = p.Exec(ctx,
+		`INSERT INTO summary_actions (user_id, video_id, action, acted_at) VALUES
+		   ($1, 'vid-1', 'watched', '2026-01-12T18:00:00Z'),
+		   ($1, 'vid-2', 'saved',   '2026-01-19T18:00:00Z')`, userA)
+	require.NoError(t, err)
+
+	got, err := s.ActiveWeeks(ctx)
+	require.NoError(t, err)
+	require.Len(t, got, 2, "both identity users must appear")
+
+	require.Equal(t, userA, got[0].UserID, "most-active user first")
+	require.Equal(t, "Ada", got[0].DisplayName)
+	require.Equal(t, 3, got[0].ActiveWeeks, "3 distinct weeks across reads+acts, shared week deduped")
+
+	require.Equal(t, userB, got[1].UserID)
+	require.Equal(t, 1, got[1].ActiveWeeks, "single read = 1 distinct week (below gate)")
+}
+
+// TestActiveWeeksEmptyWhenNoUsers: no identities → no rows (not an error).
+func TestActiveWeeksEmptyWhenNoUsers(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	got, err := s.ActiveWeeks(ctx)
+	require.NoError(t, err)
+	require.Empty(t, got)
+}

internal/adapters/store/rls_test.go

@@ -24,7 +24,7 @@ import (
 // userIsolatedTables are the tables that carry a user_id and whose policy keys
 // directly off the tapir.current_user_id GUC.
 var userIsolatedTables = []string{
-	"users", "videos", "transcripts", "summaries", "summary_actions", "video_connections",
+	"users", "videos", "transcripts", "summaries", "summary_actions", "login_events", "video_connections",
 }
 
 // allIsolatedTables adds sink_deliveries, whose ownership is derived from its
@@ -69,6 +69,10 @@ func seedUser(t *testing.T, p *pgxpool.Pool, userID string) seeded {
 		 VALUES ($1, $2, 'watched')`, userID, videoID)
 	require.NoError(t, err)
 
+	_, err = p.Exec(ctx,
+		`INSERT INTO login_events (user_id) VALUES ($1)`, userID)
+	require.NoError(t, err)
+
 	_, err = p.Exec(ctx,
 		`INSERT INTO sink_deliveries (summary_id, sink, status)
 		 VALUES ($1, 'store', 'delivered')`, summaryID)
@@ -181,13 +185,16 @@ func TestRLSEnforcesPerUserIsolation(t *testing.T) {
 	}{
 		{"update users", `UPDATE users SET display_name = 'hacked' WHERE id = $1`, b.userID},
 		{"update videos", `UPDATE videos SET title = 'hacked' WHERE user_id = $1`, b.userID},
+		{"queue videos summarize", `UPDATE videos SET summarize_requested = TRUE WHERE id = $1`, b.videoID},
 		{"update transcripts", `UPDATE transcripts SET content = 'hacked' WHERE user_id = $1`, b.userID},
 		{"update summaries", `UPDATE summaries SET summary = 'hacked' WHERE user_id = $1`, b.userID},
 		{"update summary_actions", `UPDATE summary_actions SET action = 'skipped' WHERE user_id = $1`, b.userID},
+		{"update login_events", `UPDATE login_events SET seen_at = NOW() WHERE user_id = $1`, b.userID},
 		{"update sink_deliveries", `UPDATE sink_deliveries SET status = 'hacked' WHERE summary_id = $1`, b.summaryID},
 		{"update video_connections", `UPDATE video_connections SET token_ref = 'hacked' WHERE user_id = $1`, b.userID},
 		{"delete summaries", `DELETE FROM summaries WHERE user_id = $1`, b.userID},
 		{"delete summary_actions", `DELETE FROM summary_actions WHERE user_id = $1`, b.userID},
+		{"delete login_events", `DELETE FROM login_events WHERE user_id = $1`, b.userID},
 		{"delete sink_deliveries", `DELETE FROM sink_deliveries WHERE summary_id = $1`, b.summaryID},
 		{"delete video_connections", `DELETE FROM video_connections WHERE user_id = $1`, b.userID},
 	}
@@ -204,19 +211,27 @@ func TestRLSEnforcesPerUserIsolation(t *testing.T) {
 		`SELECT summary FROM summaries WHERE user_id = $1`, b.userID).Scan(&bSummary))
 	require.Equal(t, "sum", bSummary, "B's summary must be untouched by A's writes")
 
-	var bSummaries, bActions, bDeliveries, bConnections int
+	var bSummaries, bActions, bLogins, bDeliveries, bConnections int
 	require.NoError(t, super.QueryRow(ctx,
 		`SELECT count(*) FROM summaries WHERE user_id = $1`, b.userID).Scan(&bSummaries))
 	require.NoError(t, super.QueryRow(ctx,
 		`SELECT count(*) FROM summary_actions WHERE user_id = $1`, b.userID).Scan(&bActions))
+	require.NoError(t, super.QueryRow(ctx,
+		`SELECT count(*) FROM login_events WHERE user_id = $1`, b.userID).Scan(&bLogins))
 	require.NoError(t, super.QueryRow(ctx,
 		fmt.Sprintf(`SELECT count(*) FROM sink_deliveries WHERE summary_id = '%s'`, b.summaryID)).Scan(&bDeliveries))
 	require.NoError(t, super.QueryRow(ctx,
 		`SELECT count(*) FROM video_connections WHERE user_id = $1 AND token_ref <> 'hacked'`, b.userID).Scan(&bConnections))
 	require.Equal(t, 1, bSummaries, "A's DELETE must not have removed B's summary")
 	require.Equal(t, 1, bActions, "A's DELETE must not have removed B's action")
+	require.Equal(t, 1, bLogins, "A's DELETE must not have removed B's login event")
 	require.Equal(t, 1, bDeliveries, "A's DELETE must not have removed B's delivery")
 	require.Equal(t, 1, bConnections, "A's writes must not have touched B's connection")
 
+	var bRequested bool
+	require.NoError(t, super.QueryRow(ctx,
+		`SELECT summarize_requested FROM videos WHERE user_id = $1`, b.userID).Scan(&bRequested))
+	require.False(t, bRequested, "A scoped must not have queued B's video for summarization")
+
 	_ = a // a's ids are seeded for the symmetric read assertions above
 }

internal/adapters/store/store_test.go

@@ -74,7 +74,7 @@ func rawPool(t *testing.T) *pgxpool.Pool {
 func resetDB(t *testing.T, p *pgxpool.Pool) {
 	t.Helper()
 	_, err := p.Exec(context.Background(),
-		`TRUNCATE summary_actions, sink_deliveries, summaries, transcripts, videos, users CASCADE`)
+		`TRUNCATE login_events, summary_actions, sink_deliveries, summaries, transcripts, videos, users CASCADE`)
 	require.NoError(t, err)
 }
 

internal/adapters/store/summarize_mode.go

@@ -0,0 +1,112 @@
+package store
+
+import (
+	"context"
+	"errors"
+	"fmt"
+
+	"github.com/jackc/pgx/v5"
+)
+
+// SetAutoSummarize sets the user's auto/manual summarization mode. TRUE =
+// automatic (every new video is summarized by `tapir run`); FALSE = manual (the
+// user queues videos individually). Per-user, not global (ADR-012). Scoped via
+// withUser, so RLS confines the UPDATE to the calling user's own row.
+func (s *Store) SetAutoSummarize(ctx context.Context, userID string, enabled bool) error {
+	return s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		// Ensure the row exists (FK/identity target) before the UPDATE — mirrors
+		// the Deliver/UpsertVideo paths, so toggling mode works even before the
+		// first summary lands.
+		if _, err := tx.Exec(ctx,
+			`INSERT INTO users (id) VALUES ($1) ON CONFLICT (id) DO NOTHING`,
+			userID); err != nil {
+			return fmt.Errorf("store: upsert user: %w", err)
+		}
+		if _, err := tx.Exec(ctx,
+			`UPDATE users SET auto_summarize = $1 WHERE id = $2`, enabled, userID); err != nil {
+			return fmt.Errorf("store: set auto summarize: %w", err)
+		}
+		return nil
+	})
+}
+
+// GetAutoSummarize reports the user's summarization mode (TRUE = automatic). An
+// absent user row reads as FALSE (manual), the safe default. Scoped via withUser.
+func (s *Store) GetAutoSummarize(ctx context.Context, userID string) (bool, error) {
+	var enabled bool
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		err := tx.QueryRow(ctx,
+			`SELECT auto_summarize FROM users WHERE id = $1`, userID).Scan(&enabled)
+		if errors.Is(err, pgx.ErrNoRows) {
+			enabled = false
+			return nil
+		}
+		return err
+	}); err != nil {
+		return false, fmt.Errorf("store: get auto summarize: %w", err)
+	}
+	return enabled, nil
+}
+
+// RequestSummarize queues a single video for manual summarization by setting its
+// summarize_requested flag. The next `tapir run` picks it up and clears the flag.
+// Returns ErrNotFound when the video does not exist or is not owned by the user
+// (RLS hides another user's row, so the UPDATE matches zero rows). Scoped via
+// withUser.
+func (s *Store) RequestSummarize(ctx context.Context, userID, videoID string) error {
+	return s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		ct, err := tx.Exec(ctx,
+			`UPDATE videos SET summarize_requested = TRUE WHERE id = $1`, videoID)
+		if err != nil {
+			return fmt.Errorf("store: request summarize: %w", err)
+		}
+		if ct.RowsAffected() == 0 {
+			return ErrNotFound
+		}
+		return nil
+	})
+}
+
+// RequestedVideoIDs returns the set of the user's video ids currently flagged for
+// manual summarization. The run loop loads it once per pass (mirroring
+// SeenVideoIDs) to decide which discovered videos to process in manual mode.
+// Scoped by user_id.
+func (s *Store) RequestedVideoIDs(ctx context.Context, userID string) (map[string]bool, error) {
+	requested := make(map[string]bool)
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		rows, err := tx.Query(ctx,
+			`SELECT id FROM videos WHERE user_id = $1 AND summarize_requested = TRUE`, userID)
+		if err != nil {
+			return fmt.Errorf("store: requested video ids: %w", err)
+		}
+		defer rows.Close()
+
+		for rows.Next() {
+			var id string
+			if err := rows.Scan(&id); err != nil {
+				return fmt.Errorf("store: scan requested id: %w", err)
+			}
+			requested[id] = true
+		}
+		if err := rows.Err(); err != nil {
+			return fmt.Errorf("store: iterate requested ids: %w", err)
+		}
+		return nil
+	}); err != nil {
+		return nil, err
+	}
+	return requested, nil
+}
+
+// ClearSummarizeRequested resets a video's manual queue flag, called by the run
+// loop after a queued video is successfully summarized so it is not re-processed
+// and the list view drops the "Queued" chip. Scoped via withUser.
+func (s *Store) ClearSummarizeRequested(ctx context.Context, userID, videoID string) error {
+	return s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		if _, err := tx.Exec(ctx,
+			`UPDATE videos SET summarize_requested = FALSE WHERE id = $1`, videoID); err != nil {
+			return fmt.Errorf("store: clear summarize requested: %w", err)
+		}
+		return nil
+	})
+}

internal/adapters/store/summarize_mode_test.go

@@ -0,0 +1,181 @@
+package store_test
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/jackc/pgx/v5/pgxpool"
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+)
+
+// seedBareVideo inserts a videos row with no summary, so the all-videos read and
+// the manual-queue flag can be exercised without a delivered summary.
+func seedBareVideo(t *testing.T, p *pgxpool.Pool, userID, videoID, title string) {
+	t.Helper()
+	_, err := p.Exec(context.Background(),
+		`INSERT INTO users (id) VALUES ($1) ON CONFLICT (id) DO NOTHING`, userID)
+	require.NoError(t, err)
+	_, err = p.Exec(context.Background(),
+		`INSERT INTO videos (id, user_id, provider, provider_video_id, title)
+		 VALUES ($1, $2, 'youtube', $3, $4)`,
+		videoID, userID, "pv-"+videoID[:8], title)
+	require.NoError(t, err)
+}
+
+func TestAutoSummarizeRoundTripDefaultsFalse(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	// Unknown / fresh user defaults to manual (false).
+	got, err := s.GetAutoSummarize(ctx, userA)
+	require.NoError(t, err)
+	require.False(t, got, "default mode is manual")
+
+	require.NoError(t, s.SetAutoSummarize(ctx, userA, true))
+	got, err = s.GetAutoSummarize(ctx, userA)
+	require.NoError(t, err)
+	require.True(t, got, "set to automatic round-trips")
+
+	require.NoError(t, s.SetAutoSummarize(ctx, userA, false))
+	got, err = s.GetAutoSummarize(ctx, userA)
+	require.NoError(t, err)
+	require.False(t, got, "set back to manual round-trips")
+}
+
+func TestRegisteredUserDefaultsAutoSummarizeOn(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	// A real registration creates the users row, so the column default (migration
+	// 011: TRUE) drives the mode — onboarded friends get auto out of the box.
+	id, err := s.RegisterUser(ctx, subjectA, "Alice")
+	require.NoError(t, err)
+
+	got, err := s.GetAutoSummarize(ctx, id)
+	require.NoError(t, err)
+	require.True(t, got, "new registrations default to auto-summarize (ADR-018)")
+
+	// The account-page toggle still works: a user can switch to manual.
+	require.NoError(t, s.SetAutoSummarize(ctx, id, false))
+	got, err = s.GetAutoSummarize(ctx, id)
+	require.NoError(t, err)
+	require.False(t, got, "the manual toggle still flips it off")
+}
+
+func TestRequestSummarizeSetsFlag(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	seedBareVideo(t, p, userA, videoX, "X Title")
+
+	require.NoError(t, s.RequestSummarize(ctx, userA, videoX))
+
+	requested, err := s.RequestedVideoIDs(ctx, userA)
+	require.NoError(t, err)
+	require.Equal(t, map[string]bool{videoX: true}, requested)
+
+	// Clearing drops it from the requested set.
+	require.NoError(t, s.ClearSummarizeRequested(ctx, userA, videoX))
+	requested, err = s.RequestedVideoIDs(ctx, userA)
+	require.NoError(t, err)
+	require.Empty(t, requested)
+}
+
+func TestRequestSummarizeMissingVideo(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	err := s.RequestSummarize(ctx, userA, videoX)
+	require.ErrorIs(t, err, store.ErrNotFound, "queuing a non-existent video reports not found")
+}
+
+func TestListVideosReturnsSummarizedAndUnsummarized(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	p := rawPool(t)
+	resetDB(t, p)
+
+	// videoX: discovered AND summarized. videoY: discovered, not yet summarized.
+	seedBareVideo(t, p, userA, videoX, "Summarized One")
+	seedBareVideo(t, p, userA, videoY, "Pending One")
+	require.NoError(t, s.Deliver(ctx, summary(userA, videoX, "body x")))
+	require.NoError(t, s.RequestSummarize(ctx, userA, videoY))
+
+	rows, err := s.ListVideos(ctx, userA, 50)
+	require.NoError(t, err)
+	require.Len(t, rows, 2, "both summarized and unsummarized videos are listed")
+
+	byID := map[string]store.SummaryRow{}
+	for _, r := range rows {
+		byID[r.VideoID] = r
+	}
+
+	require.True(t, byID[videoX].Summarized)
+	require.Equal(t, "body x", byID[videoX].Summary)
+	require.False(t, byID[videoX].SummarizeRequested)
+
+	require.False(t, byID[videoY].Summarized, "no summary -> Summarized false")
+	require.Empty(t, byID[videoY].Summary, "unsummarized row has empty summary")
+	require.True(t, byID[videoY].SummarizeRequested, "queued video carries the flag")
+}
+
+// TestListVideosOrderedByPublishedDescNullsLast: summarized videos sort first
+// (regardless of their date), then unsummarized by published_at DESC with
+// undated (NULL) videos last — the recency-aligned list order (UX review B2).
+func TestListVideosOrderedByPublishedDescNullsLast(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	p := rawPool(t)
+	resetDB(t, p)
+
+	const (
+		vSummOld = "cccccccc-cccc-cccc-cccc-cccccccccccc" // summarized, oldest date
+		vNewer   = "dddddddd-dddd-dddd-dddd-dddddddddddd" // unsummarized, newest
+		vOlder   = "eeeeeeee-eeee-eeee-eeee-eeeeeeeeeeee" // unsummarized, older
+		vUndated = "ffffffff-ffff-ffff-ffff-ffffffffffff" // unsummarized, no date
+	)
+	// Deliver first so the userA row exists (the videos FK needs it); the
+	// summarized video carries the OLDEST date yet must still sort first because
+	// it is summarized, proving summarized-first dominates the date sort.
+	require.NoError(t, s.Deliver(ctx, summary(userA, vSummOld, "body")))
+	seedVideo(t, p, userA, vSummOld, "Summarized Old", "youtube", "https://s", time.Date(2025, 1, 1, 0, 0, 0, 0, time.UTC))
+	seedVideo(t, p, userA, vNewer, "Newer", "youtube", "https://n", time.Date(2026, 3, 1, 0, 0, 0, 0, time.UTC))
+	seedVideo(t, p, userA, vOlder, "Older", "youtube", "https://o", time.Date(2026, 1, 1, 0, 0, 0, 0, time.UTC))
+	seedVideo(t, p, userA, vUndated, "Undated", "youtube", "https://u", time.Time{})
+
+	rows, err := s.ListVideos(ctx, userA, 50)
+	require.NoError(t, err)
+	require.Len(t, rows, 4)
+
+	got := []string{rows[0].VideoID, rows[1].VideoID, rows[2].VideoID, rows[3].VideoID}
+	require.Equal(t, []string{vSummOld, vNewer, vOlder, vUndated}, got,
+		"summarized first, then published_at DESC, NULL dates last")
+}
+
+func TestListVideosIsUserScoped(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	seedBareVideo(t, p, userA, videoX, "A only")
+
+	rows, err := s.ListVideos(ctx, userB, 50)
+	require.NoError(t, err)
+	require.Empty(t, rows, "user B must not see user A's videos")
+}
+
+func TestGetVideoRowNotFound(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	_, err := s.GetVideoRow(ctx, userA, videoX)
+	require.ErrorIs(t, err, store.ErrNotFound)
+}

internal/adapters/store/transcript_status.go

@@ -0,0 +1,102 @@
+package store
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"time"
+
+	"github.com/jackc/pgx/v5"
+)
+
+// validTranscriptStatuses bounds SetTranscriptStatus input. "" clears the status
+// (column NULL); the three named states mirror migration 007's documented values.
+var validTranscriptStatuses = map[string]bool{
+	"":             true,
+	"none":         true,
+	"rate_limited": true,
+	"fetched":      true,
+}
+
+// SetTranscriptStatus records the outcome of the last transcript attempt for a
+// video (migration 007). When status is "rate_limited" it also stamps
+// rate_limited_at = NOW() so the runner can back off; every other status clears
+// that timestamp. "" unsets the status (column NULL). An unknown status is
+// rejected. Scoped via withUser, so RLS confines the UPDATE to the caller's own
+// video; ErrNotFound when the user has no such video.
+func (s *Store) SetTranscriptStatus(ctx context.Context, userID, videoID, status string) error {
+	if !validTranscriptStatuses[status] {
+		return fmt.Errorf("store: invalid transcript status %q", status)
+	}
+	return s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		ct, err := tx.Exec(ctx,
+			`UPDATE videos
+			    SET transcript_status = NULLIF($1, ''),
+			        rate_limited_at = CASE WHEN $1 = 'rate_limited' THEN NOW() ELSE NULL END
+			  WHERE id = $2`,
+			status, videoID)
+		if err != nil {
+			return fmt.Errorf("store: set transcript status: %w", err)
+		}
+		if ct.RowsAffected() == 0 {
+			return ErrNotFound
+		}
+		return nil
+	})
+}
+
+// GetTranscriptStatus returns a video's transcript_status ("" when unset/NULL).
+// Returns ErrNotFound when the user has no such video. Scoped via withUser.
+func (s *Store) GetTranscriptStatus(ctx context.Context, userID, videoID string) (string, error) {
+	var status string
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		err := tx.QueryRow(ctx,
+			`SELECT COALESCE(transcript_status, '') FROM videos WHERE id = $1`, videoID).Scan(&status)
+		if errors.Is(err, pgx.ErrNoRows) {
+			return ErrNotFound
+		}
+		return err
+	}); err != nil {
+		if errors.Is(err, ErrNotFound) {
+			return "", ErrNotFound
+		}
+		return "", fmt.Errorf("store: get transcript status: %w", err)
+	}
+	return status, nil
+}
+
+// RateLimitedVideoIDs returns the user's videos currently in the "rate_limited"
+// state, mapped to when the 429 was stamped (rate_limited_at). The run loop loads
+// it once per pass (mirroring SeenVideoIDs) to skip re-fetching a video still
+// inside the backoff window, saving caption requests. Scoped by user_id.
+func (s *Store) RateLimitedVideoIDs(ctx context.Context, userID string) (map[string]time.Time, error) {
+	out := make(map[string]time.Time)
+	if err := s.withUser(ctx, userID, func(tx pgx.Tx) error {
+		rows, err := tx.Query(ctx,
+			`SELECT id, rate_limited_at FROM videos
+			  WHERE user_id = $1 AND transcript_status = 'rate_limited' AND rate_limited_at IS NOT NULL`,
+			userID)
+		if err != nil {
+			return fmt.Errorf("store: rate limited video ids: %w", err)
+		}
+		defer rows.Close()
+
+		for rows.Next() {
+			var (
+				id string
+				at time.Time
+			)
+			if err := rows.Scan(&id, &at); err != nil {
+				return fmt.Errorf("store: scan rate limited id: %w", err)
+			}
+			out[id] = at
+		}
+		if err := rows.Err(); err != nil {
+			return fmt.Errorf("store: iterate rate limited ids: %w", err)
+		}
+		return nil
+	}); err != nil {
+		return nil, err
+	}
+	return out, nil
+}

internal/adapters/store/transcript_status_test.go

@@ -0,0 +1,80 @@
+package store_test
+
+import (
+	"context"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+)
+
+func TestSetTranscriptStatus_RoundTrip(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	id, err := s.UpsertVideo(ctx, ytVideo(userA, "rt12345", "round trip"))
+	require.NoError(t, err)
+
+	// Unset by default.
+	got, err := s.GetTranscriptStatus(ctx, userA, id)
+	require.NoError(t, err)
+	require.Equal(t, "", got)
+
+	for _, status := range []string{"none", "fetched", "rate_limited", ""} {
+		require.NoError(t, s.SetTranscriptStatus(ctx, userA, id, status))
+		got, err := s.GetTranscriptStatus(ctx, userA, id)
+		require.NoError(t, err)
+		require.Equal(t, status, got)
+	}
+}
+
+func TestSetTranscriptStatus_RejectsInvalid(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	id, err := s.UpsertVideo(ctx, ytVideo(userA, "bad12345", "bad status"))
+	require.NoError(t, err)
+
+	require.Error(t, s.SetTranscriptStatus(ctx, userA, id, "bogus"))
+
+	// The rejected write left the status untouched.
+	got, err := s.GetTranscriptStatus(ctx, userA, id)
+	require.NoError(t, err)
+	require.Equal(t, "", got)
+}
+
+func TestSetTranscriptStatus_NotFound(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	require.ErrorIs(t, s.SetTranscriptStatus(ctx, userA, videoX, "fetched"), store.ErrNotFound)
+
+	_, err := s.GetTranscriptStatus(ctx, userA, videoX)
+	require.ErrorIs(t, err, store.ErrNotFound)
+}
+
+func TestRateLimitedVideoIDs_StampsAndClears(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	id, err := s.UpsertVideo(ctx, ytVideo(userA, "rl12345", "rate limited"))
+	require.NoError(t, err)
+
+	// Marking rate_limited stamps rate_limited_at, so the video appears.
+	require.NoError(t, s.SetTranscriptStatus(ctx, userA, id, "rate_limited"))
+	rl, err := s.RateLimitedVideoIDs(ctx, userA)
+	require.NoError(t, err)
+	require.Contains(t, rl, id)
+	require.False(t, rl[id].IsZero(), "rate_limited_at must be stamped")
+
+	// Moving off rate_limited clears the timestamp, so it drops out.
+	require.NoError(t, s.SetTranscriptStatus(ctx, userA, id, "fetched"))
+	rl, err = s.RateLimitedVideoIDs(ctx, userA)
+	require.NoError(t, err)
+	require.NotContains(t, rl, id)
+}

internal/adapters/youtube/captions.go

@@ -60,6 +60,12 @@ func (a *Adapter) FetchTranscript(ctx context.Context, v domain.Video) (domain.T
 	if err != nil {
 		return domain.Transcript{}, fmt.Errorf("download caption track for %q: %w", v.ProviderVideoID, err)
 	}
+	if status == http.StatusTooManyRequests {
+		// 429 means the IP is rate-limited; record for retry, not a permanent
+		// absence. Degrade gracefully (no error, no text) like SourceNone, but
+		// flag it distinctly so the runner backs off and retries (ADR-007/010).
+		return domain.Transcript{VideoID: v.ID, UserID: v.UserID, Source: domain.SourceRateLimited}, nil
+	}
 	if status != http.StatusOK {
 		// Owner-only 403, region/age gate, or transient unavailability: not an error.
 		return noTranscript(v), nil
@@ -229,6 +235,15 @@ func (a *Adapter) httpDo(ctx context.Context, client *http.Client, method, url s
 	for k, v := range headers {
 		req.Header.Set(k, v)
 	}
+	// Process-wide rate gate (ADR-014 item 2): every live caption fetch — player,
+	// watch-page, and timedtext baseUrl — passes the shared per-egress-IP limiter
+	// so the scheduler and the click-path cannot collectively trip 429s. Skipped
+	// when a.transport is set (the test seam) so fakes are not throttled.
+	if a.transport == nil {
+		if err := WaitFetchGate(ctx); err != nil {
+			return nil, 0, fmt.Errorf("fetch gate %s %s: %w", method, url, err)
+		}
+	}
 	resp, err := client.Do(req)
 	if err != nil {
 		return nil, 0, fmt.Errorf("%s %s: %w", method, url, err)

internal/adapters/youtube/gate.go

@@ -0,0 +1,37 @@
+package youtube
+
+import (
+	"context"
+	"time"
+
+	"golang.org/x/time/rate"
+)
+
+// globalFetchGate is the process-wide rate limiter for outbound timedtext/caption
+// fetches. A single instance is shared by ALL Adapter instances (the scheduler's
+// per-user runners + the web click-path) so they cannot collectively exceed the
+// per-egress-IP cap. ADR-014 item 2: the gate serialises/limits concurrent caption
+// fetches regardless of how many users or goroutines are upstream. The 429 is per
+// IP, not per user — so the gate is process-wide, not per-withUser, not per-video.
+//
+// Default 2s/req (burst 1): the first fetch passes immediately, subsequent fetches
+// are spaced at least 2s apart. Production overrides via SetFetchRate from config.
+var globalFetchGate = rate.NewLimiter(rate.Every(2*time.Second), 1)
+
+// SetFetchRate replaces the process-wide gate's rate with one token per interval.
+// Call once at startup from config (TAPIR_FETCH_RATE). A non-positive interval
+// installs an unlimited gate (rate.Inf) — used in dev/tests so nothing throttles.
+func SetFetchRate(interval time.Duration) {
+	if interval <= 0 {
+		globalFetchGate = rate.NewLimiter(rate.Inf, 1)
+		return
+	}
+	globalFetchGate = rate.NewLimiter(rate.Every(interval), 1)
+}
+
+// WaitFetchGate blocks until the process-wide gate allows one timedtext fetch,
+// respecting ctx cancellation. Called from httpDo before every live outbound
+// caption fetch so the scheduler and the click-path share the same egress budget.
+func WaitFetchGate(ctx context.Context) error {
+	return globalFetchGate.Wait(ctx)
+}

internal/adapters/youtube/gate_test.go

@@ -0,0 +1,84 @@
+package youtube
+
+import (
+	"context"
+	"sync"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+	"golang.org/x/time/rate"
+)
+
+// waitOn drives a test-scoped limiter the same way WaitFetchGate drives the
+// global one, so these tests exercise the gate's behaviour without mutating the
+// process-wide gate (which would pollute sibling tests / the click-path).
+func waitOn(t *testing.T, lim *rate.Limiter) func(context.Context) error {
+	t.Helper()
+	return func(ctx context.Context) error { return lim.Wait(ctx) }
+}
+
+func TestFetchGateSerialisesConcurrentCallers(t *testing.T) {
+	const (
+		n        = 5
+		interval = 10 * time.Millisecond
+	)
+	lim := rate.NewLimiter(rate.Every(interval), 1)
+	wait := waitOn(t, lim)
+
+	var (
+		inFlight, maxInFlight atomic.Int32
+		wg                    sync.WaitGroup
+	)
+	start := time.Now()
+	for i := 0; i < n; i++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			require.NoError(t, wait(context.Background()))
+			cur := inFlight.Add(1)
+			for {
+				old := maxInFlight.Load()
+				if cur <= old || maxInFlight.CompareAndSwap(old, cur) {
+					break
+				}
+			}
+			// Hold the "critical section" briefly so overlap would be observable.
+			time.Sleep(interval / 4)
+			inFlight.Add(-1)
+		}()
+	}
+	wg.Wait()
+	elapsed := time.Since(start)
+
+	require.Equal(t, int32(1), maxInFlight.Load(),
+		"the gate must admit at most one caller per interval — no overlap")
+	require.GreaterOrEqual(t, elapsed, time.Duration(n-1)*interval,
+		"N gated callers take at least (N-1)*interval wall time")
+}
+
+func TestFetchGateRespectsContextCancellation(t *testing.T) {
+	// A slow gate (1 token/hour, burst already spent) blocks; a cancelled ctx must
+	// unblock Wait with an error rather than hang.
+	lim := rate.NewLimiter(rate.Every(time.Hour), 1)
+	require.True(t, lim.Allow(), "spend the single burst token")
+
+	ctx, cancel := context.WithCancel(context.Background())
+	cancel()
+	require.Error(t, waitOn(t, lim)(ctx), "cancelled ctx must fail Wait, not block")
+}
+
+func TestSetFetchRateZeroIsUnlimited(t *testing.T) {
+	// Snapshot and restore the global so this test does not pollute the process.
+	prev := globalFetchGate
+	t.Cleanup(func() { globalFetchGate = prev })
+
+	SetFetchRate(0)
+	require.Equal(t, rate.Inf, globalFetchGate.Limit(), "0 interval = unlimited gate")
+
+	// An unlimited gate never blocks, even back-to-back.
+	for i := 0; i < 100; i++ {
+		require.NoError(t, WaitFetchGate(context.Background()))
+	}
+}

internal/adapters/youtube/youtube.go

@@ -216,6 +216,9 @@ func (a *Adapter) NewVideos(ctx context.Context, sub domain.Subscription) ([]dom
 
 	var resp playlistItemListResponse
 	if err := a.getJSON(ctx, client, "/playlistItems", q, &resp); err != nil {
+		if isHTTP404(err) {
+			return nil, &domain.ErrChannelUnavailable{ChannelID: sub.ChannelID, ChannelTitle: sub.ChannelTitle}
+		}
 		return nil, fmt.Errorf("new videos for channel %q: %w", sub.ChannelID, err)
 	}
 
@@ -269,6 +272,12 @@ func (a *Adapter) resolveUploadsPlaylist(ctx context.Context, client *http.Clien
 	return resp.Items[0].ContentDetails.RelatedPlaylists.Uploads, nil
 }
 
+// isHTTP404 reports whether err came from a YouTube API call that returned HTTP 404.
+// getRaw encodes the status as "youtube api <path>: status 404: ...".
+func isHTTP404(err error) bool {
+	return err != nil && strings.Contains(err.Error(), "status 404")
+}
+
 // getJSON issues a GET and decodes a JSON body into out. A non-200 status is an
 // error carrying a bounded slice of the response body for diagnosis.
 func (a *Adapter) getJSON(ctx context.Context, client *http.Client, path string, q url.Values, out any) error {

internal/adapters/youtube/youtube_test.go

@@ -388,6 +388,37 @@ func TestFetchTranscriptBaseURLForbiddenDegrades(t *testing.T) {
 	}
 }
 
+// A 429 on the baseUrl fetch is the IP being rate-limited, NOT a permanent
+// absence of captions: it returns SourceRateLimited (no error, no text) so the
+// runner can record it and retry after a backoff window rather than recording a
+// false "no transcript".
+func TestFetchTranscriptRateLimitedReturnsSourceRateLimited(t *testing.T) {
+	a, _ := newTestAdapter(t, func(w http.ResponseWriter, r *http.Request) {
+		switch r.URL.Path {
+		case "/youtubei/v1/player":
+			base := "http://" + r.Host
+			_, _ = w.Write([]byte(`{"captions":{"playerCaptionsTracklistRenderer":{"captionTracks":[` +
+				`{"baseUrl":"` + base + `/api/timedtext?lang=en","languageCode":"en"}]}}}`))
+		case "/api/timedtext":
+			w.WriteHeader(http.StatusTooManyRequests)
+		}
+	})
+
+	tr, err := a.FetchTranscript(context.Background(), domain.Video{ID: "v1", UserID: "u1", ProviderVideoID: "vid1"})
+	if err != nil {
+		t.Fatalf("429 on baseUrl must degrade, not error: %v", err)
+	}
+	if tr.Source != domain.SourceRateLimited {
+		t.Fatalf("expected SourceRateLimited on 429, got %q", tr.Source)
+	}
+	if tr.HasText() {
+		t.Error("expected HasText() false for SourceRateLimited")
+	}
+	if tr.Content != "" {
+		t.Errorf("expected empty content on 429, got %q", tr.Content)
+	}
+}
+
 // An empty baseUrl on the selected track degrades to SourceNone, never an error.
 func TestFetchTranscriptEmptyBaseURLDegrades(t *testing.T) {
 	a, _ := newTestAdapter(t, func(w http.ResponseWriter, r *http.Request) {

internal/config/config.go

@@ -59,9 +59,39 @@ type Config struct {
 	// PollInterval, when > 0, makes `run` loop on that cadence; 0 means run once.
 	PollInterval time.Duration
 
+	// FetchBackoff is how long the run loop waits before re-fetching a transcript
+	// that previously returned HTTP 429 (rate_limited). Inside the window the video
+	// is skipped without hitting the caption endpoint, saving requests; after it
+	// expires the video is retried. Zero means "always retry" (no backoff).
+	FetchBackoff time.Duration
+
+	// FetchRate is the minimum interval between outbound caption fetches across the
+	// whole process — the shared per-egress-IP rate gate (ADR-014 item 2). It is
+	// the gate that makes auto-summarize-on-a-schedule safe: scheduler runners and
+	// the web click-path serialise through it. Zero = unlimited (dev/tests).
+	FetchRate time.Duration
+
+	// AutoSummarizeWindow bounds auto-summarization to recent videos: in automatic
+	// mode the scheduler only summarizes videos published within this window of now.
+	// Older videos are still discovered and listed, but wait for an explicit manual
+	// "Summarize" — so a large back-catalogue does not self-inflict 429s against the
+	// caption rate gate. Zero disables the bound (summarize every unseen video, the
+	// pre-recency behaviour). Default ~7 days.
+	AutoSummarizeWindow time.Duration
+
+	// DiscoveryInterval, when > 0, makes `serve` run in-process scheduled discovery
+	// for ALL users on that cadence (ADR-018). Zero/unset = disabled, so dev and
+	// tests never auto-fetch. Single-replica assumption — see cmdServe.
+	DiscoveryInterval time.Duration
+
 	// HTTPAddr is the listen address for `tapir serve` (the Stage-0 web UI).
 	HTTPAddr string
 
+	// PublicURL is the externally-reachable base URL of the deployed service,
+	// e.g. "https://tapir.d-ma.be". Used to build absolute links handed to humans
+	// (the `tapir invite` URL). No trailing slash is assumed — callers trim it.
+	PublicURL string
+
 	// Dex OIDC (web login, ADR-011/012). When OIDCIssuer is empty, `serve` falls
 	// back to the allow-all StubAuth (local dev). When set, serve uses Dex: any
 	// Dex-authenticated subject may sign in, then registers a tapir user (ADR-012).
@@ -85,6 +115,10 @@ const (
 	defaultYTConnectRedirectURL = "https://tapir.d-ma.be/oauth/youtube/callback"
 	defaultOAuthRedirectAddr    = "localhost:8080"
 	defaultHTTPAddr             = ":8080"
+	defaultFetchBackoff         = time.Hour
+	defaultFetchRate            = 2 * time.Second
+	defaultPublicURL            = "https://tapir.d-ma.be"
+	defaultAutoSummarizeWindow  = 7 * 24 * time.Hour
 )
 
 // Load reads the environment into a Config, applying defaults. It does not
@@ -105,6 +139,7 @@ func Load() (Config, error) {
 		SecretsFile:          envOr("TAPIR_SECRETS_FILE", defaultSecretsFile()),
 		OAuthRedirectAddr:    envOr("TAPIR_OAUTH_REDIRECT_ADDR", defaultOAuthRedirectAddr),
 		HTTPAddr:             envOr("TAPIR_HTTP_ADDR", defaultHTTPAddr),
+		PublicURL:            envOr("TAPIR_PUBLIC_URL", defaultPublicURL),
 		OIDCIssuer:           os.Getenv("TAPIR_OIDC_ISSUER"),
 		DexClientID:          os.Getenv("TAPIR_DEX_CLIENT_ID"),
 		DexClientSecret:      os.Getenv("TAPIR_DEX_CLIENT_SECRET"),
@@ -124,6 +159,30 @@ func Load() (Config, error) {
 	}
 	c.PollInterval = interval
 
+	backoff, err := durationOr("TAPIR_FETCH_BACKOFF", defaultFetchBackoff)
+	if err != nil {
+		return Config{}, err
+	}
+	c.FetchBackoff = backoff
+
+	fetchRate, err := durationOr("TAPIR_FETCH_RATE", defaultFetchRate)
+	if err != nil {
+		return Config{}, err
+	}
+	c.FetchRate = fetchRate
+
+	discovery, err := durationOr("TAPIR_DISCOVERY_INTERVAL", 0)
+	if err != nil {
+		return Config{}, err
+	}
+	c.DiscoveryInterval = discovery
+
+	autoWindow, err := durationOr("TAPIR_AUTO_SUMMARIZE_WINDOW", defaultAutoSummarizeWindow)
+	if err != nil {
+		return Config{}, err
+	}
+	c.AutoSummarizeWindow = autoWindow
+
 	return c, nil
 }
 

internal/config/config_test.go

@@ -45,6 +45,12 @@ func TestLoad_AppliesDefaults(t *testing.T) {
 	if c.PollInterval != 0 {
 		t.Errorf("PollInterval = %v, want 0 (run once)", c.PollInterval)
 	}
+	if c.FetchBackoff != defaultFetchBackoff {
+		t.Errorf("FetchBackoff = %v, want default %v", c.FetchBackoff, defaultFetchBackoff)
+	}
+	if c.AutoSummarizeWindow != defaultAutoSummarizeWindow {
+		t.Errorf("AutoSummarizeWindow = %v, want default %v", c.AutoSummarizeWindow, defaultAutoSummarizeWindow)
+	}
 }
 
 func TestLoad_ParsesValues(t *testing.T) {
@@ -56,6 +62,8 @@ func TestLoad_ParsesValues(t *testing.T) {
 		"TAPIR_SUMMARIZER_TIMEOUT":    "90s",
 		"TAPIR_DB_DSN":                "postgres://x",
 		"TAPIR_POLL_INTERVAL":         "10m",
+		"TAPIR_FETCH_BACKOFF":         "30m",
+		"TAPIR_AUTO_SUMMARIZE_WINDOW": "48h",
 	})
 
 	c, err := Load()
@@ -77,6 +85,12 @@ func TestLoad_ParsesValues(t *testing.T) {
 	if c.PollInterval != 10*time.Minute {
 		t.Errorf("PollInterval = %v, want 10m", c.PollInterval)
 	}
+	if c.FetchBackoff != 30*time.Minute {
+		t.Errorf("FetchBackoff = %v, want 30m", c.FetchBackoff)
+	}
+	if c.AutoSummarizeWindow != 48*time.Hour {
+		t.Errorf("AutoSummarizeWindow = %v, want 48h", c.AutoSummarizeWindow)
+	}
 }
 
 func TestLoad_RejectsBadDuration(t *testing.T) {

internal/domain/domain.go

@@ -2,7 +2,22 @@
 // the standard library — no providers, no storage, no AI. See docs/data-model.md.
 package domain
 
-import "time"
+import (
+	"fmt"
+	"time"
+)
+
+// ErrChannelUnavailable is returned by a VideoSource when a channel's upload
+// playlist returns HTTP 404 — the channel was deleted or made private. The runner
+// stores these so the account page can surface them to the user.
+type ErrChannelUnavailable struct {
+	ChannelID    string
+	ChannelTitle string
+}
+
+func (e *ErrChannelUnavailable) Error() string {
+	return fmt.Sprintf("channel %q (%s) unavailable: playlist not found", e.ChannelTitle, e.ChannelID)
+}
 
 // Provider identifies a video platform.
 type Provider string
@@ -18,6 +33,12 @@ type TranscriptSource string
 const (
 	SourceCaptions TranscriptSource = "captions"
 	SourceNone     TranscriptSource = "none"
+	// SourceRateLimited records that the caption endpoint returned HTTP 429.
+	// Unlike SourceNone (a permanent absence), this is a transient "retry later":
+	// the IP is rate-limited, not the video caption-less. It carries no text
+	// (HasText is false), so the engine degrades the same as SourceNone, but the
+	// runner persists it distinctly to retry after a backoff window.
+	SourceRateLimited TranscriptSource = "rate_limited"
 )
 
 // User is the Tapir-side profile. At Stage 0 there is exactly one.

internal/runner/runner.go

@@ -10,11 +10,13 @@
 package runner
 
 import (
+	"cmp"
 	"context"
 	"errors"
 	"fmt"
 	"log/slog"
 	"os"
+	"slices"
 	"time"
 
 	"gitea.d-ma.be/mathias/tapir/internal/domain"
@@ -22,11 +24,56 @@ import (
 	"gitea.d-ma.be/mathias/tapir/internal/usecase"
 )
 
+// passCandidate is a video that passed all pre-filters (seen/manual/backoff)
+// and is queued for transcript fetch + summarization in this pass.
+type passCandidate struct {
+	v   domain.Video
+	pos int // discovery position — used as a stable tiebreak when published_at ties
+}
+
+// compareNewestFirst orders candidates by published_at descending, NULLS LAST,
+// with pos ascending as a stable tiebreak. Videos with a zero published_at
+// (schema 001: nullable) sort after all dated videos regardless of pos.
+func compareNewestFirst(a, b passCandidate) int {
+	aNull := a.v.PublishedAt.IsZero()
+	bNull := b.v.PublishedAt.IsZero()
+	switch {
+	case aNull && bNull:
+		return cmp.Compare(a.pos, b.pos)
+	case aNull:
+		return 1 // a is null → after b
+	case bNull:
+		return -1 // b is null → after a
+	}
+	if !a.v.PublishedAt.Equal(b.v.PublishedAt) {
+		if a.v.PublishedAt.After(b.v.PublishedAt) {
+			return -1 // newer first
+		}
+		return 1
+	}
+	return cmp.Compare(a.pos, b.pos) // same timestamp: preserve discovery order
+}
+
 // VideoStore is the durable persistence the run loop needs: assign a stable id +
-// metadata, and read the already-summarized set. *store.Store satisfies it.
+// metadata, read the already-summarized set, and (for manual summarization mode)
+// read the user's mode + queued videos and clear a video's queue flag once it has
+// been summarized. *store.Store satisfies it.
 type VideoStore interface {
 	UpsertVideo(ctx context.Context, v domain.Video) (string, error)
 	SeenVideoIDs(ctx context.Context, userID string) (map[string]bool, error)
+	GetAutoSummarize(ctx context.Context, userID string) (bool, error)
+	RequestedVideoIDs(ctx context.Context, userID string) (map[string]bool, error)
+	ClearSummarizeRequested(ctx context.Context, userID, videoID string) error
+	// RateLimitedVideoIDs maps the user's still-throttled videos to when they were
+	// rate-limited, so the loop can back off without re-hitting the caption endpoint.
+	RateLimitedVideoIDs(ctx context.Context, userID string) (map[string]time.Time, error)
+	// SetTranscriptStatus records the outcome of a transcript attempt: "none",
+	// "rate_limited" (stamps the backoff clock), or "fetched".
+	SetTranscriptStatus(ctx context.Context, userID, videoID, status string) error
+	// UpsertChannelError records a channel that returned HTTP 404 (deleted/private).
+	// Called when NewVideos returns domain.ErrChannelUnavailable; best-effort, errors
+	// are logged and never abort the pass.
+	UpsertChannelError(ctx context.Context, userID, channelID, channelTitle string) error
 }
 
 // Processor runs the core use case for a single video. *usecase.Engine
@@ -43,14 +90,43 @@ type Runner struct {
 	engine     Processor
 	userID     string
 	log        *slog.Logger
+	backoff    time.Duration    // rate-limit retry window; 0 = always retry
+	autoWindow time.Duration    // recency bound for auto-summarize; 0 = no bound
+	now        func() time.Time // injectable clock (tests); defaults to time.Now
 }
 
+// Option configures a Runner at construction. Variadic so existing call sites
+// stay valid as new knobs (backoff, clock) are added.
+type Option func(*Runner)
+
+// WithBackoff sets the rate-limit retry window. A video that returned HTTP 429 is
+// skipped (no caption fetch) until this much time has passed; 0 = always retry.
+func WithBackoff(d time.Duration) Option { return func(r *Runner) { r.backoff = d } }
+
+// WithClock overrides the clock used for backoff comparisons. Tests inject a
+// fixed time; production leaves the time.Now default.
+func WithClock(now func() time.Time) Option { return func(r *Runner) { r.now = now } }
+
+// WithAutoWindow bounds auto-summarization to videos published within d of now.
+// In auto mode a video older than d is discovered and listed but not summarized
+// automatically — it waits for an explicit manual request — so a large
+// back-catalogue does not self-inflict 429s. An explicitly requested video
+// bypasses the bound. 0 (the default) disables it (summarize every unseen video).
+func WithAutoWindow(d time.Duration) Option { return func(r *Runner) { r.autoWindow = d } }
+
 // New builds a Runner. A nil logger falls back to slog.Default.
-func New(src ports.VideoSource, store VideoStore, engine Processor, userID string, log *slog.Logger) *Runner {
+func New(src ports.VideoSource, store VideoStore, engine Processor, userID string, log *slog.Logger, opts ...Option) *Runner {
 	if log == nil {
 		log = slog.Default()
 	}
-	return &Runner{src: src, store: store, engine: engine, userID: userID, log: log}
+	r := &Runner{src: src, store: store, engine: engine, userID: userID, log: log, now: time.Now}
+	for _, opt := range opts {
+		opt(r)
+	}
+	if r.now == nil {
+		r.now = time.Now
+	}
+	return r
 }
 
 // Stats summarizes one RunOnce pass.
@@ -59,22 +135,49 @@ type Stats struct {
 	Summarized         int
 	SkippedSeen        int
 	SkippedNoText      int
+	SkippedManual      int // discovered but not queued, in manual mode
+	SkippedTooOld      int // auto mode: published outside the recency window (not requested)
+	SkippedRateLimited int // 429'd previously and still inside the backoff window
 	Errors             int
+	ChannelUnavailable int // channels that returned HTTP 404 (deleted/private)
 }
 
-// RunOnce performs a single pass over the user's subscriptions. Per-item errors
+// tooOld reports whether a video published at publishedAt falls outside the
-// are logged and collected (one bad video or channel does not abort the pass)
+// auto-summarize recency window. A zero window disables the bound, and a zero
-// and returned joined alongside the Stats gathered.
+// publishedAt (undated video) is never aged out — it cannot be dated, so it is
+// processed rather than silently stranded.
+func (r *Runner) tooOld(publishedAt time.Time) bool {
+	if r.autoWindow <= 0 || publishedAt.IsZero() {
+		return false
+	}
+	return r.now().Sub(publishedAt) > r.autoWindow
+}
+
+// RunOnce performs a single pass over the user's subscriptions in three phases:
+//
+//  1. Discovery: walk all channels, persist each candidate video (UpsertVideo),
+//     apply pre-filters (seen/manual/backoff) — same as before.
+//  2. Sort: order the surviving candidates newest-first (published_at DESC, NULLS
+//     LAST) so new users get summaries of their most recent, relevant videos first;
+//     the back-catalogue fills in behind across subsequent passes.
+//  3. Process: feed candidates to the engine in sorted order through the shared
+//     globalFetchGate — the gate is unchanged and still governs honest rate pacing.
+//
+// All existing behaviour is preserved: per-item failure isolation, the rate-limit
+// backoff skip, manual mode, channel-unavailable handling, and stats accounting.
+// Only the processing order changes within a pass.
 func (r *Runner) RunOnce(ctx context.Context) (Stats, error) {
 	var (
 		stats      Stats
 		errs       []error
+		candidates []passCandidate
+		pos        int
 	)
 
 	// Throttle transcript fetches: unauthenticated caption scraping gets
 	// soft-throttled by YouTube under heavy back-to-back volume (captionTracks
 	// silently stripped from the player response). A small per-video delay keeps
-	// a full pass under the radar. TAPIR_FETCH_DELAY (Go duration), 0 = off.
+	// the fetch rate polite. TAPIR_FETCH_DELAY (Go duration), 0 = off.
 	fetchDelay, _ := time.ParseDuration(os.Getenv("TAPIR_FETCH_DELAY"))
 
 	seen, err := r.store.SeenVideoIDs(ctx, r.userID)
@@ -82,21 +185,62 @@ func (r *Runner) RunOnce(ctx context.Context) (Stats, error) {
 		return stats, fmt.Errorf("runner: load seen videos: %w", err)
 	}
 
+	// Summarization mode (per-user, ADR-012). Auto = summarize every unseen video.
+	// Manual = discover/persist videos (visible in list) but only summarize ones
+	// explicitly queued via the web UI (summarize_requested).
+	auto, err := r.store.GetAutoSummarize(ctx, r.userID)
+	if err != nil {
+		return stats, fmt.Errorf("runner: load summarize mode: %w", err)
+	}
+	// requested is needed in manual mode (the queue) and in auto mode when a
+	// recency window is active (an explicit request bypasses the bound).
+	var requested map[string]bool
+	if !auto || r.autoWindow > 0 {
+		requested, err = r.store.RequestedVideoIDs(ctx, r.userID)
+		if err != nil {
+			return stats, fmt.Errorf("runner: load requested videos: %w", err)
+		}
+	}
+
+	// Rate-limit backoff: videos that 429'd on a prior pass, mapped to when.
+	// Within the backoff window they are skipped before any caption fetch.
+	// Disabled when backoff <= 0 ("always retry").
+	var rateLimited map[string]time.Time
+	if r.backoff > 0 {
+		rateLimited, err = r.store.RateLimitedVideoIDs(ctx, r.userID)
+		if err != nil {
+			return stats, fmt.Errorf("runner: load rate-limited videos: %w", err)
+		}
+	}
+
 	subs, err := r.src.ListSubscriptions(ctx, r.userID)
 	if err != nil {
 		return stats, fmt.Errorf("runner: list subscriptions: %w", err)
 	}
 
+	// ── Phase 1: discover, persist, filter ───────────────────────────────────
+	// Walk all channels. Persist every video (UpsertVideo) so it appears in the
+	// list regardless of whether it will be summarized this pass. Apply pre-filters
+	// and collect surviving candidates with their discovery position.
 	for _, sub := range subs {
 		vids, err := r.src.NewVideos(ctx, sub)
 		if err != nil {
+			var unavail *domain.ErrChannelUnavailable
+			if errors.As(err, &unavail) {
+				stats.ChannelUnavailable++
+				r.log.Warn("channel unavailable (playlist 404)", "channel", sub.ChannelTitle, "channel_id", sub.ChannelID)
+				if storeErr := r.store.UpsertChannelError(ctx, r.userID, unavail.ChannelID, unavail.ChannelTitle); storeErr != nil {
+					r.log.Warn("failed to store channel error", "err", storeErr)
+				}
+			} else {
 				errs = append(errs, fmt.Errorf("new videos for %q: %w", sub.ChannelTitle, err))
 				stats.Errors++
+			}
 			continue
 		}
 		for _, v := range vids {
 			stats.Candidates++
-			v.UserID = r.userID // keep the dedup/FK key consistent with config
+			v.UserID = r.userID
 
 			id, err := r.store.UpsertVideo(ctx, v)
 			if err != nil {
@@ -110,26 +254,89 @@ func (r *Runner) RunOnce(ctx context.Context) (Stats, error) {
 				stats.SkippedSeen++
 				continue
 			}
-			seen[id] = true // also guard against the same video within this pass
+			seen[id] = true // guard against duplicates within this pass
 
+			// Manual mode: skip summarization for videos the user has not queued.
+			// Discovery already happened (UpsertVideo above), so the video is
+			// visible in the list; it just isn't summarized until requested.
+			if !auto && !requested[id] {
+				stats.SkippedManual++
+				continue
+			}
+
+			// Recency bound (auto mode): summarize only recent videos automatically;
+			// older ones are discovered + listed (UpsertVideo above) but wait for an
+			// explicit manual request, so a large back-catalogue does not self-inflict
+			// 429s against the caption rate gate. A requested video bypasses the bound.
+			if auto && !requested[id] && r.tooOld(v.PublishedAt) {
+				stats.SkippedTooOld++
+				continue
+			}
+
+			// Still inside the rate-limit backoff window: skip without fetching.
+			if at, ok := rateLimited[id]; ok && r.now().Sub(at) < r.backoff {
+				stats.SkippedRateLimited++
+				r.log.Info("skipped video (rate-limited, backing off)", "video", v.ProviderVideoID, "title", v.Title)
+				continue
+			}
+
+			candidates = append(candidates, passCandidate{v: v, pos: pos})
+			pos++
+		}
+	}
+
+	// ── Phase 2: sort newest-first, NULLS LAST ────────────────────────────────
+	// Within this pass, process the newest videos first so a new user gets
+	// summaries of their most recent content quickly; the back-catalogue fills in
+	// behind across subsequent passes. Both this background batch and the foreground
+	// "Try now" button honour the shared globalFetchGate — ordering is onboarding
+	// prioritisation, not rate-limit evasion.
+	slices.SortStableFunc(candidates, compareNewestFirst)
+
+	// ── Phase 3: process in sorted order ─────────────────────────────────────
+	for _, c := range candidates {
 		if fetchDelay > 0 {
 			time.Sleep(fetchDelay)
 		}
-			res, err := r.engine.ProcessNewVideo(ctx, v)
+		res, err := r.engine.ProcessNewVideo(ctx, c.v)
 		if err != nil {
-				errs = append(errs, fmt.Errorf("process %q: %w", v.ProviderVideoID, err))
+			errs = append(errs, fmt.Errorf("process %q: %w", c.v.ProviderVideoID, err))
 			stats.Errors++
 			continue
 		}
+		id := c.v.ID
 		switch {
+		case res.Skipped && res.TranscriptSource == string(domain.SourceRateLimited):
+			// Fresh 429: stamp the backoff clock so the next pass skips it.
+			stats.SkippedRateLimited++
+			if err := r.store.SetTranscriptStatus(ctx, r.userID, id, "rate_limited"); err != nil {
+				errs = append(errs, fmt.Errorf("set rate_limited status %q: %w", c.v.ProviderVideoID, err))
+				stats.Errors++
+			}
+			r.log.Info("skipped video (rate-limited)", "video", c.v.ProviderVideoID, "title", c.v.Title)
 		case res.Skipped:
 			stats.SkippedNoText++
-				r.log.Info("skipped video (no transcript)", "video", v.ProviderVideoID, "title", v.Title)
+			if err := r.store.SetTranscriptStatus(ctx, r.userID, id, "none"); err != nil {
+				errs = append(errs, fmt.Errorf("set none status %q: %w", c.v.ProviderVideoID, err))
+				stats.Errors++
+			}
+			r.log.Info("skipped video (no transcript)", "video", c.v.ProviderVideoID, "title", c.v.Title)
 		case res.Summary != nil:
 			stats.Summarized++
-				r.log.Info("summarized video", "video", v.ProviderVideoID, "title", v.Title,
+			if err := r.store.SetTranscriptStatus(ctx, r.userID, id, "fetched"); err != nil {
-					"provider", res.Summary.AIProvider, "model", res.Summary.AIModel)
+				errs = append(errs, fmt.Errorf("set fetched status %q: %w", c.v.ProviderVideoID, err))
+				stats.Errors++
 			}
+			// In manual mode the video was explicitly queued; clear the flag so
+			// it is not re-summarized and the UI drops the "Queued" chip.
+			if !auto {
+				if err := r.store.ClearSummarizeRequested(ctx, r.userID, id); err != nil {
+					errs = append(errs, fmt.Errorf("clear summarize flag %q: %w", c.v.ProviderVideoID, err))
+					stats.Errors++
+				}
+			}
+			r.log.Info("summarized video", "video", c.v.ProviderVideoID, "title", c.v.Title,
+				"provider", res.Summary.AIProvider, "model", res.Summary.AIModel)
 		}
 	}
 
@@ -145,7 +352,9 @@ func (r *Runner) Loop(ctx context.Context, interval time.Duration) error {
 		r.log.Info("run pass complete",
 			"candidates", stats.Candidates, "summarized", stats.Summarized,
 			"skipped_seen", stats.SkippedSeen, "skipped_no_text", stats.SkippedNoText,
-			"errors", stats.Errors)
+			"skipped_manual", stats.SkippedManual, "skipped_too_old", stats.SkippedTooOld,
+			"skipped_rate_limited", stats.SkippedRateLimited,
+			"channel_unavailable", stats.ChannelUnavailable, "errors", stats.Errors)
 		if err != nil {
 			r.log.Warn("run pass had errors", "err", err)
 		}

internal/runner/runner_test.go

@@ -5,6 +5,7 @@ import (
 	"io"
 	"log/slog"
 	"testing"
+	"time"
 
 	"github.com/stretchr/testify/require"
 
@@ -40,9 +41,16 @@ func (f *fakeSource) FetchTranscript(_ context.Context, v domain.Video) (domain.
 
 // fakeStore assigns deterministic ids ("id-"+provider video id) so a pre-seeded
 // seen set lines up with UpsertVideo output, modelling cross-restart dedup.
+// auto controls the summarization mode; requested is the manual-mode queue keyed
+// by store id; cleared records the ids whose queue flag the runner reset.
 type fakeStore struct {
 	seen        map[string]bool
 	upserted    []domain.Video
+	auto        bool
+	requested   map[string]bool
+	cleared     []string
+	rateLimited map[string]time.Time // id -> when 429'd (seeds the backoff window)
+	statuses    map[string]string    // id -> last SetTranscriptStatus value
 }
 
 func (f *fakeStore) UpsertVideo(_ context.Context, v domain.Video) (string, error) {
@@ -58,6 +66,41 @@ func (f *fakeStore) SeenVideoIDs(_ context.Context, _ string) (map[string]bool,
 	return cp, nil
 }
 
+func (f *fakeStore) GetAutoSummarize(_ context.Context, _ string) (bool, error) {
+	return f.auto, nil
+}
+
+func (f *fakeStore) RequestedVideoIDs(_ context.Context, _ string) (map[string]bool, error) {
+	cp := make(map[string]bool, len(f.requested))
+	for k, v := range f.requested {
+		cp[k] = v
+	}
+	return cp, nil
+}
+
+func (f *fakeStore) ClearSummarizeRequested(_ context.Context, _, videoID string) error {
+	f.cleared = append(f.cleared, videoID)
+	return nil
+}
+
+func (f *fakeStore) RateLimitedVideoIDs(_ context.Context, _ string) (map[string]time.Time, error) {
+	cp := make(map[string]time.Time, len(f.rateLimited))
+	for k, v := range f.rateLimited {
+		cp[k] = v
+	}
+	return cp, nil
+}
+
+func (f *fakeStore) UpsertChannelError(_ context.Context, _, _, _ string) error { return nil }
+
+func (f *fakeStore) SetTranscriptStatus(_ context.Context, _, videoID, status string) error {
+	if f.statuses == nil {
+		f.statuses = map[string]string{}
+	}
+	f.statuses[videoID] = status
+	return nil
+}
+
 type fakeSummarizer struct{}
 
 func (fakeSummarizer) Summarize(_ context.Context, v domain.Video, _ domain.Transcript) (domain.Summary, error) {
@@ -80,6 +123,12 @@ func vid(provID, title string) domain.Video {
 	return domain.Video{UserID: testUser, Provider: domain.ProviderYouTube, ProviderVideoID: provID, Title: title}
 }
 
+func vidAt(provID, title string, publishedAt time.Time) domain.Video {
+	v := vid(provID, title)
+	v.PublishedAt = publishedAt
+	return v
+}
+
 func quietLogger() *slog.Logger {
 	return slog.New(slog.NewTextHandler(io.Discard, nil))
 }
@@ -91,7 +140,7 @@ func TestRunOnce_SummarizesNewVideos(t *testing.T) {
 		subs:   []domain.Subscription{sub("chan1", "Channel One")},
 		videos: map[string][]domain.Video{"chan1": {vid("v1", "Video 1"), vid("v2", "Video 2")}},
 	}
-	st := &fakeStore{seen: map[string]bool{}}
+	st := &fakeStore{seen: map[string]bool{}, auto: true}
 	sink := &recordingSink{}
 	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
 	r := runner.New(src, st, eng, testUser, quietLogger())
@@ -114,7 +163,7 @@ func TestRunOnce_SkipsAlreadySummarized(t *testing.T) {
 		videos: map[string][]domain.Video{"chan1": {vid("v1", "Video 1"), vid("v2", "Video 2")}},
 	}
 	// v1 was summarized in a prior run (durable seen set).
-	st := &fakeStore{seen: map[string]bool{"id-v1": true}}
+	st := &fakeStore{seen: map[string]bool{"id-v1": true}, auto: true}
 	sink := &recordingSink{}
 	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
 	r := runner.New(src, st, eng, testUser, quietLogger())
@@ -133,7 +182,7 @@ func TestRunOnce_SkipsVideosWithoutTranscript(t *testing.T) {
 		videos:      map[string][]domain.Video{"chan1": {vid("v1", "Video 1")}},
 		transcripts: map[string]domain.Transcript{"v1": {Source: domain.SourceNone}},
 	}
-	st := &fakeStore{seen: map[string]bool{}}
+	st := &fakeStore{seen: map[string]bool{}, auto: true}
 	sink := &recordingSink{}
 	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
 	r := runner.New(src, st, eng, testUser, quietLogger())
@@ -145,13 +194,199 @@ func TestRunOnce_SkipsVideosWithoutTranscript(t *testing.T) {
 	require.Empty(t, sink.delivered, "no summary delivered when there is no transcript")
 }
 
+func TestRunOnce_ManualMode_SkipsUnrequested(t *testing.T) {
+	src := &fakeSource{
+		subs:   []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {vid("v1", "Video 1"), vid("v2", "Video 2")}},
+	}
+	// Manual mode, nothing queued: discover (upsert) but summarize nothing.
+	st := &fakeStore{seen: map[string]bool{}, auto: false, requested: map[string]bool{}}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger())
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 2, stats.Candidates)
+	require.Equal(t, 2, stats.SkippedManual, "manual mode skips unqueued videos")
+	require.Equal(t, 0, stats.Summarized)
+	require.Empty(t, sink.delivered, "no summary in manual mode without a request")
+	require.Len(t, st.upserted, 2, "discovery still persists every candidate")
+}
+
+func TestRunOnce_ManualMode_ProcessesRequested(t *testing.T) {
+	src := &fakeSource{
+		subs:   []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {vid("v1", "Video 1"), vid("v2", "Video 2")}},
+	}
+	// Manual mode, v1 queued (by store id). Only v1 is summarized; its flag clears.
+	st := &fakeStore{seen: map[string]bool{}, auto: false, requested: map[string]bool{"id-v1": true}}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger())
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 1, stats.Summarized, "only the queued video is summarized")
+	require.Equal(t, 1, stats.SkippedManual, "the unqueued video is skipped")
+	require.Len(t, sink.delivered, 1)
+	require.Equal(t, "id-v1", sink.delivered[0].VideoID)
+	require.Equal(t, []string{"id-v1"}, st.cleared, "the queue flag is cleared after summarizing")
+}
+
+// --- recency window (B1) ---------------------------------------------------
+
+// TestRunOnce_AutoMode_SkipsOldVideos: with a recency window set, auto mode
+// summarizes only videos published within the window; older ones are discovered
+// (upserted) but not auto-summarized — they wait for a manual request.
+func TestRunOnce_AutoMode_SkipsOldVideos(t *testing.T) {
+	base := time.Date(2026, 6, 8, 12, 0, 0, 0, time.UTC)
+	src := &fakeSource{
+		subs: []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {
+			vidAt("recent", "Recent", base.Add(-24*time.Hour)), // 1d old → in window
+			vidAt("old", "Old", base.Add(-30*24*time.Hour)),    // 30d old → out of window
+		}},
+	}
+	st := &fakeStore{seen: map[string]bool{}, auto: true}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger(),
+		runner.WithAutoWindow(7*24*time.Hour), runner.WithClock(func() time.Time { return base }))
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 1, stats.Summarized, "only the recent video is auto-summarized")
+	require.Equal(t, 1, stats.SkippedTooOld, "the old video is skipped by the recency bound")
+	require.Len(t, sink.delivered, 1)
+	require.Equal(t, "id-recent", sink.delivered[0].VideoID)
+	require.Len(t, st.upserted, 2, "both videos are still discovered and listed")
+}
+
+// TestRunOnce_AutoMode_OldVideoRequestedBypassesWindow: an explicit manual
+// request (summarize_requested) overrides the recency bound even in auto mode.
+func TestRunOnce_AutoMode_OldVideoRequestedBypassesWindow(t *testing.T) {
+	base := time.Date(2026, 6, 8, 12, 0, 0, 0, time.UTC)
+	src := &fakeSource{
+		subs:   []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {vidAt("old", "Old", base.Add(-30*24*time.Hour))}},
+	}
+	st := &fakeStore{seen: map[string]bool{}, auto: true, requested: map[string]bool{"id-old": true}}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger(),
+		runner.WithAutoWindow(7*24*time.Hour), runner.WithClock(func() time.Time { return base }))
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 1, stats.Summarized, "a requested old video is summarized despite the window")
+	require.Equal(t, 0, stats.SkippedTooOld)
+	require.Len(t, sink.delivered, 1)
+}
+
+// TestRunOnce_AutoWindowZero_SummarizesOld: a zero window disables the bound —
+// the pre-recency behaviour (summarize every unseen video) is preserved.
+func TestRunOnce_AutoWindowZero_SummarizesOld(t *testing.T) {
+	base := time.Date(2026, 6, 8, 12, 0, 0, 0, time.UTC)
+	src := &fakeSource{
+		subs:   []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {vidAt("old", "Old", base.Add(-365*24*time.Hour))}},
+	}
+	st := &fakeStore{seen: map[string]bool{}, auto: true}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger(),
+		runner.WithClock(func() time.Time { return base })) // no WithAutoWindow → 0
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 1, stats.Summarized, "window disabled → old video summarized")
+	require.Equal(t, 0, stats.SkippedTooOld)
+}
+
+// TestRunOnce_AutoMode_UndatedVideoSummarized: a video with no published_at
+// cannot be aged out — it is processed, not silently stranded.
+func TestRunOnce_AutoMode_UndatedVideoSummarized(t *testing.T) {
+	base := time.Date(2026, 6, 8, 12, 0, 0, 0, time.UTC)
+	src := &fakeSource{
+		subs:   []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {vid("undated", "Undated")}}, // zero PublishedAt
+	}
+	st := &fakeStore{seen: map[string]bool{}, auto: true}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger(),
+		runner.WithAutoWindow(7*24*time.Hour), runner.WithClock(func() time.Time { return base }))
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 1, stats.Summarized, "an undated video is processed, not aged out")
+	require.Equal(t, 0, stats.SkippedTooOld)
+}
+
+// noFetchSource fails the test if a transcript fetch happens — used to prove the
+// runner skips a rate-limited video before touching the caption endpoint.
+type noFetchSource struct{ *fakeSource }
+
+func (noFetchSource) FetchTranscript(context.Context, domain.Video) (domain.Transcript, error) {
+	panic("FetchTranscript must not be called for a rate-limited video within the backoff window")
+}
+
+func TestRunOnce_SkipsRateLimitedWithinBackoff(t *testing.T) {
+	base := time.Date(2026, 6, 3, 12, 0, 0, 0, time.UTC)
+	src := &fakeSource{
+		subs:   []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {vid("v1", "Video 1")}},
+	}
+	// v1 was rate-limited 5m ago; backoff is 1h, so it is still inside the window.
+	st := &fakeStore{
+		seen:        map[string]bool{},
+		auto:        true,
+		rateLimited: map[string]time.Time{"id-v1": base.Add(-5 * time.Minute)},
+	}
+	eng := usecase.NewEngine(noFetchSource{src}, fakeSummarizer{}, &recordingSink{})
+	r := runner.New(noFetchSource{src}, st, eng, testUser, quietLogger(),
+		runner.WithBackoff(time.Hour), runner.WithClock(func() time.Time { return base }))
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 1, stats.SkippedRateLimited, "still throttled -> skipped")
+	require.Equal(t, 0, stats.Summarized)
+	require.Empty(t, st.statuses, "no status write: the engine was never invoked")
+}
+
+func TestRunOnce_RetriesRateLimitedAfterBackoff(t *testing.T) {
+	base := time.Date(2026, 6, 3, 12, 0, 0, 0, time.UTC)
+	src := &fakeSource{
+		subs:   []domain.Subscription{sub("chan1", "Channel One")},
+		videos: map[string][]domain.Video{"chan1": {vid("v1", "Video 1")}},
+	}
+	// v1 was rate-limited 2h ago; backoff is 1h, so the window has expired.
+	st := &fakeStore{
+		seen:        map[string]bool{},
+		auto:        true,
+		rateLimited: map[string]time.Time{"id-v1": base.Add(-2 * time.Hour)},
+	}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger(),
+		runner.WithBackoff(time.Hour), runner.WithClock(func() time.Time { return base }))
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 0, stats.SkippedRateLimited, "window expired -> not skipped")
+	require.Equal(t, 1, stats.Summarized, "the video is retried and summarized")
+	require.Len(t, sink.delivered, 1)
+	require.Equal(t, "fetched", st.statuses["id-v1"], "status advances to fetched on success")
+}
+
 func TestRunOnce_UpsertsEveryCandidate(t *testing.T) {
 	src := &fakeSource{
 		subs:   []domain.Subscription{sub("chan1", "Channel One")},
 		videos: map[string][]domain.Video{"chan1": {vid("v1", "Video 1"), vid("v2", "Video 2")}},
 	}
 	// Even an already-seen video gets upserted so its metadata stays fresh.
-	st := &fakeStore{seen: map[string]bool{"id-v1": true}}
+	st := &fakeStore{seen: map[string]bool{"id-v1": true}, auto: true}
 	eng := usecase.NewEngine(src, fakeSummarizer{}, &recordingSink{})
 	r := runner.New(src, st, eng, testUser, quietLogger())
 
@@ -159,3 +394,89 @@ func TestRunOnce_UpsertsEveryCandidate(t *testing.T) {
 	require.NoError(t, err)
 	require.Len(t, st.upserted, 2, "every candidate is upserted, including seen ones")
 }
+
+// TestRunOnce_NewestFirstOrdering asserts that within a pass, candidates are
+// processed newest-first (published_at DESC, NULLS LAST) across all channels,
+// and that the set of processed videos is identical to what per-channel inline
+// processing would produce (only the order differs).
+//
+// Fixture: two channels, four videos with mixed published_at (one NULL).
+//
+//	Per-channel (before): chanA=[v-old, v-mid], chanB=[v-new, v-null]
+//	                    → [v-old, v-mid, v-new, v-null]
+//	Newest-first (after): [v-new, v-mid, v-old, v-null]
+func TestRunOnce_NewestFirstOrdering(t *testing.T) {
+	old := time.Date(2024, 1, 1, 0, 0, 0, 0, time.UTC)
+	mid := time.Date(2024, 6, 1, 0, 0, 0, 0, time.UTC)
+	newt := time.Date(2024, 12, 1, 0, 0, 0, 0, time.UTC)
+	// zero time = NULL published_at (schema 001: nullable)
+
+	src := &fakeSource{
+		subs: []domain.Subscription{
+			sub("chanA", "Channel A"),
+			sub("chanB", "Channel B"),
+		},
+		videos: map[string][]domain.Video{
+			"chanA": {
+				vidAt("v-old", "Old Video", old),
+				vidAt("v-mid", "Mid Video", mid),
+			},
+			"chanB": {
+				vidAt("v-new", "New Video", newt),
+				vidAt("v-null", "No Date Video", time.Time{}), // NULL
+			},
+		},
+	}
+	st := &fakeStore{seen: map[string]bool{}, auto: true}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger())
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+
+	// Same set: all 4 candidates processed regardless of order.
+	require.Equal(t, 4, stats.Candidates)
+	require.Equal(t, 4, stats.Summarized, "same set of videos processed as per-channel order")
+	require.Len(t, sink.delivered, 4)
+
+	// Build video-id → delivery-position map.
+	order := make(map[string]int, len(sink.delivered))
+	for i, s := range sink.delivered {
+		order[s.VideoID] = i
+		t.Logf("position %d: %s", i, s.VideoID)
+	}
+
+	require.Less(t, order["id-v-new"], order["id-v-mid"], "newest (Dec) before mid (Jun)")
+	require.Less(t, order["id-v-mid"], order["id-v-old"], "mid (Jun) before old (Jan)")
+	require.Less(t, order["id-v-old"], order["id-v-null"], "dated before NULL (NULLS LAST)")
+}
+
+// TestRunOnce_NewestFirstNullsOnly asserts that when all candidates have NULL
+// published_at, discovery order (stable) is preserved as the tiebreak.
+func TestRunOnce_NewestFirstNullsOnly(t *testing.T) {
+	src := &fakeSource{
+		subs: []domain.Subscription{
+			sub("chanA", "Channel A"),
+			sub("chanB", "Channel B"),
+		},
+		videos: map[string][]domain.Video{
+			"chanA": {vidAt("v1", "V1", time.Time{}), vidAt("v2", "V2", time.Time{})},
+			"chanB": {vidAt("v3", "V3", time.Time{})},
+		},
+	}
+	st := &fakeStore{seen: map[string]bool{}, auto: true}
+	sink := &recordingSink{}
+	eng := usecase.NewEngine(src, fakeSummarizer{}, sink)
+	r := runner.New(src, st, eng, testUser, quietLogger())
+
+	stats, err := r.RunOnce(context.Background())
+	require.NoError(t, err)
+	require.Equal(t, 3, stats.Summarized, "all null-date videos processed")
+
+	// Discovery order: chanA[v1, v2], chanB[v3] → [v1, v2, v3].
+	// All have NULL published_at so the sort is stable; discovery order must hold.
+	require.Equal(t, "id-v1", sink.delivered[0].VideoID)
+	require.Equal(t, "id-v2", sink.delivered[1].VideoID)
+	require.Equal(t, "id-v3", sink.delivered[2].VideoID)
+}

internal/usecase/engine.go

@@ -45,6 +45,11 @@ type ProcessResult struct {
 	Video   domain.Video
 	Skipped bool
 	Reason  string // set when Skipped (e.g. "no transcript")
+	// TranscriptSource is how the transcript resolved (or that there was none):
+	// the domain.TranscriptSource value as a string. The runner reads it to tell a
+	// permanent absence (SourceNone) from a transient 429 (SourceRateLimited) and
+	// persist the right transcript_status. Empty when a fetch error short-circuits.
+	TranscriptSource string
 	Summary          *domain.Summary // nil when Skipped
 }
 
@@ -59,7 +64,9 @@ func (e *Engine) ProcessNewVideo(ctx context.Context, v domain.Video) (ProcessRe
 	if !t.HasText() {
 		// No usable transcript: record the skip, produce no summary, deliver nothing
 		// (captions-first, ADR-007; the watcher uses this to avoid reprocessing).
-		return ProcessResult{Video: v, Skipped: true, Reason: "no transcript"}, nil
+		// Surface the source so the runner separates SourceNone (permanent) from
+		// SourceRateLimited (retry after a backoff window).
+		return ProcessResult{Video: v, Skipped: true, Reason: "no transcript", TranscriptSource: string(t.Source)}, nil
 	}
 
 	sum, err := e.AI.Summarize(ctx, v, t)
@@ -76,7 +83,7 @@ func (e *Engine) ProcessNewVideo(ctx context.Context, v domain.Video) (ProcessRe
 		}
 	}
 
-	return ProcessResult{Video: v, Summary: &sum}, errors.Join(errs...)
+	return ProcessResult{Video: v, Summary: &sum, TranscriptSource: string(t.Source)}, errors.Join(errs...)
 }
 
 // ProcessNewVideos walks a user's subscriptions and processes each newly seen

internal/web/account.go

@@ -28,7 +28,17 @@ func (a *App) handleAccount(w http.ResponseWriter, r *http.Request) {
 	if u, ok := a.Auth.CurrentUser(r); ok {
 		email = u.Email
 	}
-	a.render(w, r, AccountPage(name, email, conns, takeFlash(w, r)))
+	auto, err := a.Store.GetAutoSummarize(r.Context(), userID)
+	if err != nil {
+		a.serverError(w, r, "summarize mode", err)
+		return
+	}
+	channelErrs, err := a.Store.ListChannelErrors(r.Context(), userID)
+	if err != nil {
+		a.serverError(w, r, "channel errors", err)
+		return
+	}
+	a.render(w, r, AccountPage(name, email, conns, auto, channelErrs, takeFlash(w, r)))
 }
 
 // handleDisconnect removes a provider connection: it deletes the OAuth token from

internal/web/auth.go

@@ -1,6 +1,7 @@
-// Package web is the Stage-0 HTTP read/write surface (ADR-011, docs/ui-spec.md).
+// Package web is the multi-user HTTP read/write surface (ADR-012, docs/ui-spec.md).
 // It serves the summary reader over the existing store; the engine and ports are
-// untouched (ADR-003).
+// untouched (ADR-003). ADR-011 shipped this as a single-user Stage-0 reader; ADR-012
+// opened Stage 1 — multiple Dex-authenticated users with DB-enforced (RLS) isolation.
 //
 // This file defines the auth SEAM so the Dex session layer (internal/web/oidc)
 // and the page/handler layer can be built independently: handlers depend only on
@@ -10,9 +11,10 @@ package web
 
 import "net/http"
 
-// User is the authenticated principal. Subject is the Dex subject used for the
+// User is the authenticated principal. Subject is the Dex subject — the key for the
-// single-user allowlist (ADR-011); store operations key off the configured
+// user_identities lookup (ADR-012) that resolves to a tapir user_id (UUID); store
-// tapir user_id (UUID), not this subject.
+// operations scope every row by that id, not by this subject. A subject with no
+// users row is routed through the registration gate (see registration.go).
 type User struct {
 	Subject string
 	Email   string

internal/web/connect.go

@@ -23,6 +23,15 @@ type Connections interface {
 	UpsertConnection(ctx context.Context, userID string, c store.Connection) error
 }
 
+// DiscoveryTrigger requests an out-of-band discovery pass for a user. The connect
+// flow fires it the moment a YouTube account is linked so videos appear promptly
+// instead of waiting for the next scheduled pass (#6). Enqueue must be
+// non-blocking and safe to call from the request goroutine; the implementation
+// owns serialization with the scheduler (one pass at a time). nil = no trigger.
+type DiscoveryTrigger interface {
+	Enqueue(userID string)
+}
+
 // connectStateTTL bounds how long a generated CSRF state is valid between the
 // connect redirect and the provider callback.
 const connectStateTTL = 10 * time.Minute
@@ -43,6 +52,10 @@ type ConnectHandler struct {
 	Conns   Connections
 	Log     *slog.Logger
 
+	// Discovery, when set, is fired after a successful connect so the new
+	// connection's videos are discovered immediately (#6). Optional.
+	Discovery DiscoveryTrigger
+
 	states *connectStateStore
 	now    func() time.Time
 }
@@ -134,6 +147,12 @@ func (h *ConnectHandler) handleCallback(w http.ResponseWriter, r *http.Request)
 		return
 	}
 
+	// Discover this user's videos now rather than waiting for the next scheduled
+	// pass (#6). Non-blocking; the trigger serializes with the scheduler.
+	if h.Discovery != nil {
+		h.Discovery.Enqueue(userID)
+	}
+
 	setFlash(w, flashConnected)
 	http.Redirect(w, r, "/", http.StatusSeeOther)
 }

internal/web/connect_test.go

@@ -45,6 +45,24 @@ func (c *fakeConns) UpsertConnection(_ context.Context, userID string, conn stor
 	return nil
 }
 
+// fakeTrigger records Enqueue calls so a test can assert connect fired discovery.
+type fakeTrigger struct {
+	mu    sync.Mutex
+	users []string
+}
+
+func (f *fakeTrigger) Enqueue(userID string) {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	f.users = append(f.users, userID)
+}
+
+func (f *fakeTrigger) seen() []string {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	return append([]string(nil), f.users...)
+}
+
 // tokenServer fakes Google's token endpoint, returning body for any POST.
 func tokenServer(t *testing.T, body string) *httptest.Server {
 	t.Helper()
@@ -126,6 +144,36 @@ func TestCallbackExchangesAndRecordsConnection(t *testing.T) {
 	require.Equal(t, wantRef, conns.conn.TokenRef)
 }
 
+func TestCallbackTriggersDiscovery(t *testing.T) {
+	srv := tokenServer(t,
+		`{"access_token":"at","refresh_token":"rt-secret","token_type":"Bearer","expires_in":3600}`)
+	app := newConnectApp(t, srv.URL, &fakeWriter{}, &fakeConns{})
+	trig := &fakeTrigger{}
+	app.Connect.Discovery = trig
+
+	state := connectState(t, app)
+	rec := do(t, app, httptest.NewRequest(http.MethodGet,
+		"/oauth/youtube/callback?state="+state+"&code=the-code", nil))
+
+	require.Equal(t, http.StatusSeeOther, rec.Code)
+	require.Equal(t, []string{userID}, trig.seen(),
+		"a successful connect must trigger discovery for the connecting user")
+}
+
+func TestCallbackNoDiscoveryOnFailedConnect(t *testing.T) {
+	srv := tokenServer(t,
+		`{"access_token":"at","refresh_token":"rt","token_type":"Bearer","expires_in":3600}`)
+	app := newConnectApp(t, srv.URL, &fakeWriter{}, &fakeConns{})
+	trig := &fakeTrigger{}
+	app.Connect.Discovery = trig
+
+	// No state → CSRF reject → nothing connected, so no discovery.
+	rec := do(t, app, httptest.NewRequest(http.MethodGet,
+		"/oauth/youtube/callback?code=the-code", nil))
+	require.Equal(t, http.StatusBadRequest, rec.Code)
+	require.Empty(t, trig.seen(), "a failed connect must not trigger discovery")
+}
+
 func TestCallbackRejectsMissingState(t *testing.T) {
 	srv := tokenServer(t,
 		`{"access_token":"at","refresh_token":"rt","token_type":"Bearer","expires_in":3600}`)

internal/web/handlers.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"log/slog"
 	"net/http"
+	"time"
 
 	"github.com/a-h/templ"
 
@@ -16,17 +17,36 @@ import (
 // the concrete *store.Store). *store.Store satisfies it; tests can substitute a
 // fake without a database.
 type Store interface {
-	ListSummaries(ctx context.Context, userID string, limit int) ([]store.SummaryRow, error)
+	ListVideos(ctx context.Context, userID string, limit int) ([]store.SummaryRow, error)
 	GetSummaryByVideo(ctx context.Context, userID, videoID string) (*store.SummaryRow, error)
+	GetVideoRow(ctx context.Context, userID, videoID string) (*store.SummaryRow, error)
 	ActionsFor(ctx context.Context, userID string, videoIDs []string) (map[string][]string, error)
 	SetAction(ctx context.Context, userID, videoID, action string) error
 	ClearAction(ctx context.Context, userID, videoID, action string) error
 
+	// Summarization mode: the per-user auto/manual toggle and the per-video
+	// manual queue (the "Summarize" button). The runner consumes the queue.
+	GetAutoSummarize(ctx context.Context, userID string) (bool, error)
+	SetAutoSummarize(ctx context.Context, userID string, enabled bool) error
+	RequestSummarize(ctx context.Context, userID, videoID string) error
+
 	// Account management (the /account page, disconnect, delete-account).
 	ConnectionsForUser(ctx context.Context, userID string) ([]store.Connection, error)
 	DeleteConnection(ctx context.Context, userID, provider string) error
 	DeleteUser(ctx context.Context, userID string) error
 	DisplayName(ctx context.Context, userID string) (string, error)
+	// ListChannelErrors returns channels that returned HTTP 404 (deleted/private) on
+	// the most recent discovery pass, shown on the account page as a warning.
+	ListChannelErrors(ctx context.Context, userID string) ([]store.ChannelError, error)
+
+	// SetTranscriptStatus clears or updates a video's transcript backoff state.
+	// Used by handleRetryNow to clear rate_limited_at before immediate processing.
+	SetTranscriptStatus(ctx context.Context, userID, videoID, status string) error
+
+	// StampLogin records (throttled, one row per user per day) that the resolved
+	// user was active on this request — the read-side Stage-0 usage signal the
+	// registration gate stamps for every authenticated request.
+	StampLogin(ctx context.Context, userID string) error
 }
 
 // SecretRemover deletes secret material by its opaque ref. *secrets.FileStore
@@ -54,6 +74,39 @@ type App struct {
 	// Secrets removes a user's OAuth tokens on disconnect / delete-account. The
 	// account routes require it; cmd/tapir wires the file-backed store.
 	Secrets SecretRemover
+	// Processor, when non-nil, summarizes a queued video immediately in a
+	// background goroutine (the "Summarize" button kicks it off). Nil = queue-only:
+	// the button flips the DB flag and the next `tapir run` does the work.
+	Processor Processor
+	// Processing tracks in-flight immediate summarizations so the status endpoint
+	// shows the animation until the summary lands. The zero value is ready to use.
+	Processing ProcessingSet
+	// RecencyWindow mirrors the auto-summarize recency bound: un-summarized videos
+	// published before now-RecencyWindow collapse into the "older videos"
+	// disclosure on the list, so the readable summaries are not buried (B3). Zero
+	// disables the collapse (everything stays inline).
+	RecencyWindow time.Duration
+	// Now is an injectable clock for the recency cutoff (tests fix it). Nil =
+	// time.Now.
+	Now func() time.Time
+}
+
+// now returns the App's clock (time.Now unless overridden for tests).
+func (a *App) now() time.Time {
+	if a.Now != nil {
+		return a.Now()
+	}
+	return time.Now()
+}
+
+// recencyCutoff is the timestamp before which an un-summarized video counts as
+// "older" and collapses into the disclosure. A zero RecencyWindow yields the zero
+// time, which bucketRows treats as "collapse disabled".
+func (a *App) recencyCutoff() time.Time {
+	if a.RecencyWindow <= 0 {
+		return time.Time{}
+	}
+	return a.now().Add(-a.RecencyWindow)
 }
 
 func (a *App) logger() *slog.Logger {
@@ -69,6 +122,7 @@ func (a *App) logger() *slog.Logger {
 func (a *App) Router() http.Handler {
 	root := http.NewServeMux()
 	root.HandleFunc("GET /healthz", a.handleHealthz)
+	root.HandleFunc("GET /welcome", a.handleWelcome)
 	root.Handle("GET /static/", staticHandler())
 	root.Handle("/auth/", a.Auth.Routes())
 
@@ -76,6 +130,9 @@ func (a *App) Router() http.Handler {
 	app.HandleFunc("GET /{$}", a.handleList)
 	app.HandleFunc("GET /v/{videoId}", a.handleDetail)
 	app.HandleFunc("POST /v/{videoId}/action", a.handleAction)
+	app.HandleFunc("POST /v/{videoId}/summarize", a.handleRequestSummarize)
+	app.HandleFunc("POST /v/{videoId}/retry-now", a.handleRetryNow)
+	app.HandleFunc("GET /v/{videoId}/status", a.handleStatus)
 	app.HandleFunc("GET /register", a.handleRegisterForm)
 	app.HandleFunc("POST /register", a.handleRegister)
 
@@ -84,6 +141,7 @@ func (a *App) Router() http.Handler {
 	app.HandleFunc("GET /account", a.handleAccount)
 	app.HandleFunc("POST /account/disconnect/{provider}", a.handleDisconnect)
 	app.HandleFunc("POST /account/delete", a.handleDeleteAccount)
+	app.HandleFunc("POST /account/summarize-mode", a.handleSummarizeMode)
 
 	// Web-initiated YouTube connect (ADR-006). Gated like every app route, so
 	// CurrentUserID is set and the connection binds to the authenticated user.
@@ -100,6 +158,15 @@ func (a *App) Router() http.Handler {
 	return root
 }
 
+// handleWelcome renders the public landing page (/welcome). It is mounted outside
+// Auth.Middleware, so it must not assume a session: CurrentUser peeks the cookie
+// without redirecting and the page renders the logged-out or logged-in variant
+// accordingly.
+func (a *App) handleWelcome(w http.ResponseWriter, r *http.Request) {
+	user, ok := a.Auth.CurrentUser(r)
+	a.render(w, r, WelcomePage(user, ok))
+}
+
 // handleHealthz is the unauthenticated liveness/readiness probe.
 func (a *App) handleHealthz(w http.ResponseWriter, _ *http.Request) {
 	w.Header().Set("Content-Type", "text/plain; charset=utf-8")
@@ -119,20 +186,37 @@ func (a *App) handleList(w http.ResponseWriter, r *http.Request) {
 		Channel:        q.Get("channel"),
 		From:           q.Get("from"),
 		To:             q.Get("to"),
+		OnlySummarized: q.Get("summarized") == "1",
 	}
 
-	rows, err := a.Store.ListSummaries(r.Context(), userID, 0)
+	allRows, err := a.Store.ListVideos(r.Context(), userID, 0)
 	if err != nil {
-		a.serverError(w, r, "list summaries", err)
+		a.serverError(w, r, "list videos", err)
 		return
 	}
-	rows = f.apply(rows)
+	stats := pipelineStats(allRows)
+	rows := f.apply(allRows)
+	buckets := bucketRows(rows, a.recencyCutoff())
+
+	// hasConnected drives the empty state: a fresh account with a connection but
+	// no discovery pass yet has zero rows, and we want it to read "connected,
+	// summaries land gradually" rather than "nothing here". Only needed when the
+	// list is empty.
+	hasConnected := false
+	if buckets.empty() {
+		conns, err := a.Store.ConnectionsForUser(r.Context(), userID)
+		if err != nil {
+			a.serverError(w, r, "connections for user", err)
+			return
+		}
+		hasConnected = len(conns) > 0
+	}
 
 	if isHTMX(r) {
-		a.render(w, r, summaryList(rows))
+		a.render(w, r, summaryList(buckets, hasConnected))
 		return
 	}
-	a.render(w, r, ListPage(rows, f, takeFlash(w, r)))
+	a.render(w, r, ListPage(buckets, f, stats, takeFlash(w, r), hasConnected))
 }
 
 // handleDetail renders one summary in full (highlights, takeaways, action group).
@@ -199,6 +283,146 @@ func (a *App) handleAction(w http.ResponseWriter, r *http.Request) {
 	http.Redirect(w, r, "/v/"+videoID, http.StatusSeeOther)
 }
 
+// handleRequestSummarize handles the "Summarize" button. It always flips the DB
+// flag (summarize_requested) so the work is durable. With a Processor wired it
+// then summarizes immediately in the background and answers with the animated
+// processing card that polls /status until done; without one (queue-only) it
+// answers with the "Queued" card — the next `tapir run` does the work. Without
+// JS it redirects back to the list (POST→redirect→GET).
+func (a *App) handleRequestSummarize(w http.ResponseWriter, r *http.Request) {
+	userID, ok := a.currentUserID(w, r)
+	if !ok {
+		return
+	}
+	videoID := r.PathValue("videoId")
+
+	err := a.Store.RequestSummarize(r.Context(), userID, videoID)
+	if errors.Is(err, store.ErrNotFound) {
+		http.NotFound(w, r)
+		return
+	}
+	if err != nil {
+		a.serverError(w, r, "request summarize", err)
+		return
+	}
+
+	if !isHTMX(r) {
+		http.Redirect(w, r, "/", http.StatusSeeOther)
+		return
+	}
+	row, err := a.Store.GetVideoRow(r.Context(), userID, videoID)
+	if err != nil {
+		a.serverError(w, r, "get video", err)
+		return
+	}
+
+	if a.Processor != nil {
+		a.startProcessing(userID, videoID)
+		a.render(w, r, processingCard(*row))
+		return
+	}
+	a.render(w, r, VideoCard(*row))
+}
+
+// handleRetryNow handles the "Try now" button on rate-limited video cards. It
+// clears the rate_limited_at backoff so the scheduler won't skip the video, then
+// triggers an immediate ProcessVideo — same background path as handleRequestSummarize.
+// The rate gate (globalFetchGate) still applies, so this is safe under concurrent use.
+func (a *App) handleRetryNow(w http.ResponseWriter, r *http.Request) {
+	userID, ok := a.currentUserID(w, r)
+	if !ok {
+		return
+	}
+	videoID := r.PathValue("videoId")
+
+	// Clear the backoff so the scheduler won't skip this video on the next pass.
+	if err := a.Store.SetTranscriptStatus(r.Context(), userID, videoID, "none"); err != nil {
+		a.serverError(w, r, "clear rate limit", err)
+		return
+	}
+
+	if !isHTMX(r) {
+		http.Redirect(w, r, "/", http.StatusSeeOther)
+		return
+	}
+	row, err := a.Store.GetVideoRow(r.Context(), userID, videoID)
+	if err != nil {
+		a.serverError(w, r, "get video", err)
+		return
+	}
+	if a.Processor != nil {
+		a.startProcessing(userID, videoID)
+		a.render(w, r, processingCard(*row))
+		return
+	}
+	a.render(w, r, VideoCard(*row))
+}
+
+// startProcessing marks a video in-flight and summarizes it in the background.
+// The goroutine uses a detached context — not the request's, which is cancelled
+// when the handler returns — and clears the in-flight mark on completion. On
+// error the DB flag stays set, so the video remains queued for the next
+// `tapir run`; a successful Processor.ProcessVideo clears it itself.
+func (a *App) startProcessing(userID, videoID string) {
+	key := processingKey(userID, videoID)
+	a.Processing.Add(key)
+	go func() {
+		defer a.Processing.Remove(key)
+		if err := a.Processor.ProcessVideo(context.Background(), userID, videoID); err != nil {
+			a.logger().Error("background summarize", "user", userID, "video", videoID, "err", err)
+		}
+	}()
+}
+
+// handleStatus is the HTMX poll target for an in-flight summarization. It returns
+// the card in its current state: the full summary card once the summary exists,
+// otherwise the animated processing card while still in-flight (which keeps
+// polling), or the queued/button card when neither holds. VideoCard carries no
+// polling attributes, so HTMX stops polling once it swaps in.
+func (a *App) handleStatus(w http.ResponseWriter, r *http.Request) {
+	userID, ok := a.currentUserID(w, r)
+	if !ok {
+		return
+	}
+	videoID := r.PathValue("videoId")
+
+	row, err := a.Store.GetVideoRow(r.Context(), userID, videoID)
+	if errors.Is(err, store.ErrNotFound) {
+		http.NotFound(w, r)
+		return
+	}
+	if err != nil {
+		a.serverError(w, r, "get video", err)
+		return
+	}
+
+	if row.Summarized || !a.Processing.Has(processingKey(userID, videoID)) {
+		a.render(w, r, VideoCard(*row))
+		return
+	}
+	a.render(w, r, processingCard(*row))
+}
+
+// handleSummarizeMode toggles the user's auto/manual summarization mode. The form
+// submits the desired new value (enabled=true|false). For HTMX it returns the
+// refreshed mode control; without JS it redirects back to the account page.
+func (a *App) handleSummarizeMode(w http.ResponseWriter, r *http.Request) {
+	userID, ok := a.currentUserID(w, r)
+	if !ok {
+		return
+	}
+	enabled := r.FormValue("enabled") == "true"
+	if err := a.Store.SetAutoSummarize(r.Context(), userID, enabled); err != nil {
+		a.serverError(w, r, "set summarize mode", err)
+		return
+	}
+	if !isHTMX(r) {
+		http.Redirect(w, r, "/account", http.StatusSeeOther)
+		return
+	}
+	a.render(w, r, summarizeModeControl(enabled))
+}
+
 // currentUserID returns the tapir user_id the registration gate resolved for this
 // request. Behind the gate it is always present; a miss means a handler was
 // reached without scoping (a wiring bug), so it answers 500 and reports false.

internal/web/handlers_test.go

@@ -72,7 +72,7 @@ func rawPool(t *testing.T) *pgxpool.Pool {
 func truncateAll(t *testing.T, p *pgxpool.Pool) {
 	t.Helper()
 	_, err := p.Exec(context.Background(),
-		`TRUNCATE summary_actions, sink_deliveries, summaries, transcripts, videos, users CASCADE`)
+		`TRUNCATE login_events, summary_actions, sink_deliveries, summaries, transcripts, videos, users CASCADE`)
 	require.NoError(t, err)
 }
 
@@ -185,8 +185,10 @@ func TestListRendersRowsAndActionState(t *testing.T) {
 func TestListHTMXReturnsFragment(t *testing.T) {
 	ctx := context.Background()
 	app := newApp(t)
-	resetDB(t, rawPool(t))
+	p := rawPool(t)
+	resetDB(t, p)
 	require.NoError(t, deliver(ctx, app, videoX, "body x"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
 
 	req := httptest.NewRequest(http.MethodGet, "/", nil)
 	req.Header.Set("HX-Request", "true")
@@ -232,6 +234,18 @@ func TestDetailRendersHighlightsAndTakeaways(t *testing.T) {
 	require.Contains(t, html, "takeaway one")
 	require.Contains(t, html, `id="action-buttons"`, "action button group present")
 	require.Contains(t, html, "Watched")
+	require.Contains(t, html, `class="segmented"`, "watched/skipped render as a segmented control (UX review C5)")
+	require.Less(t, strings.Index(html, "Skipped"), strings.Index(html, "Saved"),
+		"Saved sits after the watched/skipped segment")
+	require.Contains(t, html, "← Summaries", "back link to the list (UX review C3)")
+
+	// The detail page leads with the attention-saving payload (UX review A8):
+	// Takeaways ("is it worth my time?") above Highlights above the full Summary.
+	takeaways := strings.Index(html, "takeaway one")
+	highlights := strings.Index(html, "highlight one")
+	summary := strings.Index(html, "the full summary body")
+	require.Less(t, takeaways, highlights, "Takeaways render before Highlights")
+	require.Less(t, highlights, summary, "Highlights render before the full Summary")
 }
 
 func TestDetailNotFound(t *testing.T) {
@@ -290,6 +304,162 @@ func TestActionRejectsUnknownVerb(t *testing.T) {
 	require.Equal(t, http.StatusBadRequest, rec.Code)
 }
 
+func TestListShowsSummarizeButtonForUnsummarized(t *testing.T) {
+	app := newApp(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	// A discovered-but-unsummarized video (no summary delivered).
+	seedVideo(t, p, videoX, "Pending Title", "https://x", time.Time{})
+
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+
+	require.Contains(t, html, "Pending Title", "unsummarized videos are listed too")
+	require.Contains(t, html, "Summarize", "a Summarize button is offered")
+	require.Contains(t, html, "/v/"+videoX+"/summarize", "button posts to the queue endpoint")
+	require.Contains(t, html, "card-pending", "muted pending treatment")
+	require.NotContains(t, html, "Queued", "not queued yet")
+}
+
+// TestListCollapsesOlderAndNoCaption verifies the feed IA (UX review B3/B4):
+// summarized + recent un-summarized cards lead inline; older un-summarized
+// videos collapse into a single disclosure; caption-less videos collapse into a
+// one-line count instead of dead cards.
+func TestListCollapsesOlderAndNoCaption(t *testing.T) {
+	ctx := context.Background()
+	app := newApp(t)
+	now := time.Date(2026, 6, 8, 12, 0, 0, 0, time.UTC)
+	app.RecencyWindow = 7 * 24 * time.Hour
+	app.Now = func() time.Time { return now }
+	p := rawPool(t)
+	resetDB(t, p)
+
+	const (
+		vRecent = "cccccccc-cccc-cccc-cccc-cccccccccccc" // 2d old, unsummarized → inline
+		vOld    = "dddddddd-dddd-dddd-dddd-dddddddddddd" // 30d old, unsummarized → disclosure
+		vNoCap  = "eeeeeeee-eeee-eeee-eeee-eeeeeeeeeeee" // caption-less → collapsed line
+	)
+	require.NoError(t, deliver(ctx, app, videoX, "body x")) // summarized, recent
+	seedVideo(t, p, videoX, "Summarized X", "https://x", now.Add(-24*time.Hour))
+	seedVideo(t, p, vRecent, "Recent Pending", "https://r", now.Add(-2*24*time.Hour))
+	seedVideo(t, p, vOld, "Old Pending", "https://o", now.Add(-30*24*time.Hour))
+	seedVideo(t, p, vNoCap, "No Caption Vid", "https://n", now.Add(-40*24*time.Hour))
+	_, err := p.Exec(ctx, `UPDATE videos SET transcript_status = 'none' WHERE id = $1`, vNoCap)
+	require.NoError(t, err)
+
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+
+	disclosure := strings.Index(html, "Show 1 older videos")
+	require.GreaterOrEqual(t, disclosure, 0, "older-videos disclosure present")
+	// Summarized + recent un-summarized lead inline, above the disclosure.
+	require.Less(t, strings.Index(html, "Summarized X"), disclosure, "summarized card is inline")
+	require.Less(t, strings.Index(html, "Recent Pending"), disclosure, "recent pending is inline")
+	// The older video is hidden inside the disclosure, after its summary.
+	require.Greater(t, strings.Index(html, "Old Pending"), disclosure, "older video lives in the disclosure")
+	// Caption-less video is a one-line count, never a card.
+	require.Contains(t, html, "have no captions")
+	require.NotContains(t, html, "No Caption Vid", "caption-less video is collapsed, not a card")
+}
+
+// TestListHidesFilterBarWhenEmpty: a genuinely empty list shows no filter bar
+// (the connect CTA stands alone), but a filter that matches nothing still shows
+// the bar so it can be cleared (UX review C1).
+func TestListHidesFilterBarWhenEmpty(t *testing.T) {
+	app := newApp(t)
+	resetDB(t, rawPool(t))
+
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.NotContains(t, body(t, rec), `class="filters"`, "no filter bar on an empty account")
+
+	rec = do(t, app, httptest.NewRequest(http.MethodGet, "/?channel=nope", nil))
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Contains(t, body(t, rec), `class="filters"`, "filtered-to-empty keeps the bar so it can be cleared")
+}
+
+func TestRequestSummarizeQueuesAndRendersCard(t *testing.T) {
+	ctx := context.Background()
+	app := newApp(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	seedVideo(t, p, videoX, "Pending Title", "https://x", time.Time{})
+
+	rec := postSummarize(t, app, videoX, true)
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+	require.Contains(t, html, "Queued", "card now shows the queued state")
+	require.NotContains(t, html, ">Summarize<", "the Summarize button is gone once queued")
+
+	// The flag is persisted, so the next run picks it up.
+	row, err := app.Store.GetVideoRow(ctx, userID, videoX)
+	require.NoError(t, err)
+	require.True(t, row.SummarizeRequested)
+}
+
+func TestRequestSummarizeNonHTMXRedirects(t *testing.T) {
+	ctx := context.Background()
+	app := newApp(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	seedVideo(t, p, videoX, "Pending Title", "https://x", time.Time{})
+
+	rec := postSummarize(t, app, videoX, false)
+	require.Equal(t, http.StatusSeeOther, rec.Code)
+	require.Equal(t, "/", rec.Header().Get("Location"))
+
+	row, err := app.Store.GetVideoRow(ctx, userID, videoX)
+	require.NoError(t, err)
+	require.True(t, row.SummarizeRequested, "queued on the no-JS path too")
+}
+
+func TestRequestSummarizeNotFound(t *testing.T) {
+	app := newApp(t)
+	resetDB(t, rawPool(t))
+	rec := postSummarize(t, app, videoX, true)
+	require.Equal(t, http.StatusNotFound, rec.Code, "queuing an unknown video is a 404")
+}
+
+func TestSummarizeModeToggle(t *testing.T) {
+	ctx := context.Background()
+	app := newApp(t)
+	resetDB(t, rawPool(t))
+
+	// Account page defaults to automatic (ADR-018: onboarded users get
+	// zero-friction discovery — the list fills and summarizes itself).
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/account", nil))
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+	require.Contains(t, html, "Automatic", "default mode shown")
+	require.Contains(t, html, "Switch to manual")
+
+	// Toggle to manual via HTMX returns the refreshed control.
+	req := httptest.NewRequest(http.MethodPost, "/account/summarize-mode",
+		strings.NewReader("enabled=false"))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("HX-Request", "true")
+	rec = do(t, app, req)
+	require.Equal(t, http.StatusOK, rec.Code)
+	html = body(t, rec)
+	require.Contains(t, html, "Manual")
+	require.Contains(t, html, "Switch to automatic")
+
+	got, err := app.Store.GetAutoSummarize(ctx, userID)
+	require.NoError(t, err)
+	require.False(t, got, "mode persisted")
+}
+
+func postSummarize(t *testing.T, app *web.App, videoID string, htmx bool) *httptest.ResponseRecorder {
+	t.Helper()
+	req := httptest.NewRequest(http.MethodPost, "/v/"+videoID+"/summarize", nil)
+	if htmx {
+		req.Header.Set("HX-Request", "true")
+	}
+	return do(t, app, req)
+}
+
 // deliver stores a summary through the App's store under test.
 func deliver(ctx context.Context, app *web.App, videoID, text string) error {
 	return app.Store.(*store.Store).Deliver(ctx, summary(videoID, text))

internal/web/oidc/oidc.go

@@ -161,11 +161,11 @@ func (d *DexAuth) Middleware(h http.Handler) http.Handler {
 		}
 		sid, ok := d.sessionID(r)
 		if !ok {
-			d.redirectToLogin(w, r)
+			d.redirectUnauthenticated(w, r)
 			return
 		}
 		if _, ok := d.sessions.get(sid, d.now()); !ok {
-			d.redirectToLogin(w, r)
+			d.redirectUnauthenticated(w, r)
 			return
 		}
 		d.sessions.refresh(sid, d.now().Add(d.sessionTTL)) // sliding refresh
@@ -266,7 +266,21 @@ func (d *DexAuth) handleLogout(w http.ResponseWriter, r *http.Request) {
 		d.sessions.delete(sid)
 	}
 	d.clearSessionCookie(w)
-	http.Redirect(w, r, loginPath, http.StatusFound)
+	// Land on the public landing page, not the login endpoint: a just-logged-out
+	// visitor should see /welcome, not be bounced straight back into a Dex login.
+	http.Redirect(w, r, "/welcome", http.StatusFound)
+}
+
+// redirectUnauthenticated sends an unauthenticated visitor somewhere useful: the
+// bare root goes to the public landing page (/welcome), any deeper guarded path
+// goes to login so the post-login round-trip can return them to it. isPublicPath
+// has already let /welcome and /auth/* through, so this never loops.
+func (d *DexAuth) redirectUnauthenticated(w http.ResponseWriter, r *http.Request) {
+	if r.URL.Path == "/" {
+		http.Redirect(w, r, "/welcome", http.StatusFound)
+		return
+	}
+	d.redirectToLogin(w, r)
 }
 
 func (d *DexAuth) redirectToLogin(w http.ResponseWriter, r *http.Request) {
@@ -305,5 +319,5 @@ func (d *DexAuth) clearSessionCookie(w http.ResponseWriter) {
 }
 
 func isPublicPath(p string) bool {
-	return p == "/healthz" || strings.HasPrefix(p, "/auth/")
+	return p == "/healthz" || p == "/welcome" || strings.HasPrefix(p, "/auth/")
 }

internal/web/oidc/oidc_test.go

@@ -248,9 +248,15 @@ func TestMiddlewareRedirectsUnauthenticated(t *testing.T) {
 		w.WriteHeader(http.StatusOK)
 	}))
 
+	// The bare root sends an unauthenticated visitor to the public landing page.
 	rec := httptest.NewRecorder()
 	guarded.ServeHTTP(rec, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, http.StatusFound, rec.Code)
+	require.Equal(t, "/welcome", rec.Header().Get("Location"))
 
+	// A deeper guarded path goes to login so the post-login round-trip returns there.
+	rec = httptest.NewRecorder()
+	guarded.ServeHTTP(rec, httptest.NewRequest(http.MethodGet, "/v/some-id", nil))
 	require.Equal(t, http.StatusFound, rec.Code)
 	require.Equal(t, "/auth/login", rec.Header().Get("Location"))
 }
@@ -280,7 +286,7 @@ func TestMiddlewarePublicPathsBypassAuth(t *testing.T) {
 		w.WriteHeader(http.StatusOK)
 	}))
 
-	for _, path := range []string{"/healthz", "/auth/login"} {
+	for _, path := range []string{"/healthz", "/welcome", "/auth/login"} {
 		rec := httptest.NewRecorder()
 		guarded.ServeHTTP(rec, httptest.NewRequest(http.MethodGet, path, nil))
 		require.Equal(t, http.StatusOK, rec.Code, "expected %s to bypass auth", path)
@@ -298,6 +304,7 @@ func TestLogoutClearsSession(t *testing.T) {
 	auth.Routes().ServeHTTP(rec, req)
 
 	require.Equal(t, http.StatusFound, rec.Code)
+	require.Equal(t, "/welcome", rec.Header().Get("Location"), "logout lands on the public page")
 	cleared := sessionCookie(t, rec.Result())
 	require.Less(t, cleared.MaxAge, 0, "logout expires the cookie")
 

internal/web/processing.go

@@ -0,0 +1,42 @@
+package web
+
+import (
+	"context"
+	"sync"
+)
+
+// Processor runs the core summarization use case for a single already-discovered
+// video — resolve its transcript, summarize, deliver to the store. *usecase.Engine
+// wrapped with the store satisfies it (wired in cmd/tapir). Optional on App: a nil
+// Processor means queue-only — the "Summarize" button only flips the DB flag and
+// the next `tapir run` does the work.
+type Processor interface {
+	ProcessVideo(ctx context.Context, userID, videoID string) error
+}
+
+// ProcessingSet tracks the (user, video) ids currently being summarized in-process
+// so the status endpoint can show the animation until the summary lands. It is
+// ephemeral (single-instance Stage-1): a restart drops it, and the DB holds the
+// durable state — the summary is present, or summarize_requested is still set so
+// `tapir run` retries. The zero value is ready to use; methods are concurrency-safe.
+type ProcessingSet struct {
+	m sync.Map
+}
+
+// Add marks a key in-flight.
+func (p *ProcessingSet) Add(key string) { p.m.Store(key, struct{}{}) }
+
+// Remove clears a key once its summarization finishes (success or failure).
+func (p *ProcessingSet) Remove(key string) { p.m.Delete(key) }
+
+// Has reports whether a key is currently in-flight.
+func (p *ProcessingSet) Has(key string) bool {
+	_, ok := p.m.Load(key)
+	return ok
+}
+
+// processingKey scopes the in-flight key by user so one user's summarization is
+// never confused with another's for the same video id.
+func processingKey(userID, videoID string) string {
+	return userID + "|" + videoID
+}

internal/web/processing_handler_test.go

@@ -0,0 +1,131 @@
+package web_test
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/web"
+)
+
+// fakeProcessor records ProcessVideo calls. With block set it parks until the
+// channel is closed, so a test can observe the handler return before the
+// background work finishes (proving it ran in a goroutine).
+type fakeProcessor struct {
+	block chan struct{}
+	done  chan struct{}
+	calls []string
+}
+
+func (f *fakeProcessor) ProcessVideo(_ context.Context, _, videoID string) error {
+	if f.block != nil {
+		<-f.block
+	}
+	f.calls = append(f.calls, videoID)
+	if f.done != nil {
+		close(f.done)
+	}
+	return nil
+}
+
+func TestRequestSummarizeImmediateProcessing(t *testing.T) {
+	app := newApp(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	seedVideo(t, p, videoX, "Pending Title", "https://x", time.Time{})
+
+	fp := &fakeProcessor{block: make(chan struct{}), done: make(chan struct{})}
+	app.Processor = fp
+
+	rec := postSummarize(t, app, videoX, true)
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+
+	// The processing card came back while ProcessVideo is still parked on block:
+	// the work runs in a goroutine, the handler did not wait for it.
+	require.Contains(t, html, "Summarizing", "processing card returned")
+	require.Contains(t, html, "╭", "charm box rendered")
+	require.Contains(t, html, "▓", "tapir body block chars rendered")
+	require.Contains(t, html, "∩", "wiggling snout frame rendered")
+	require.Contains(t, html, web.CharmPurple, "charm palette applied to the border")
+	require.Contains(t, html, "/v/"+videoX+"/status", "card polls the status endpoint")
+	require.Contains(t, html, `hx-trigger="every 2s"`, "card auto-polls every 2s")
+	require.NotContains(t, html, "Queued", "not the queue-only card")
+
+	close(fp.block)
+	select {
+	case <-fp.done:
+	case <-time.After(2 * time.Second):
+		t.Fatal("ProcessVideo was not called in the background")
+	}
+	require.Equal(t, []string{videoX}, fp.calls)
+}
+
+func TestStatusProcessingThenDone(t *testing.T) {
+	ctx := context.Background()
+	app := newApp(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	seedVideo(t, p, videoX, "Pending Title", "https://x", time.Time{})
+
+	// Park ProcessVideo so the video stays in-flight while we poll status.
+	fp := &fakeProcessor{block: make(chan struct{})}
+	app.Processor = fp
+	require.Equal(t, http.StatusOK, postSummarize(t, app, videoX, true).Code)
+
+	// Processing: status returns the animation card, still polling.
+	rec := getStatus(t, app, videoX)
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+	require.Contains(t, html, "Summarizing", "in-flight → animation card")
+	require.Contains(t, html, `hx-trigger="every 2s"`, "still polling")
+
+	close(fp.block)
+
+	// Done: once a summary exists, status returns the summary card with no poll.
+	require.NoError(t, deliver(ctx, app, videoX, "the summary body"))
+	rec = getStatus(t, app, videoX)
+	require.Equal(t, http.StatusOK, rec.Code)
+	html = body(t, rec)
+	require.NotContains(t, html, "Summarizing", "done → no animation")
+	require.NotContains(t, html, "every 2s", "done card does not poll (polling stops)")
+	require.Contains(t, html, "/v/"+videoX+"\"", "links to the detail page")
+}
+
+func TestStatusQueuedWhenNotInFlight(t *testing.T) {
+	ctx := context.Background()
+	app := newApp(t)
+	p := rawPool(t)
+	resetDB(t, p)
+	seedVideo(t, p, videoX, "Pending Title", "https://x", time.Time{})
+
+	// Flag set but nothing in-flight (e.g. queue-only, or after a restart).
+	require.NoError(t, app.Store.RequestSummarize(ctx, userID, videoX))
+
+	rec := getStatus(t, app, videoX)
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+	require.Contains(t, html, "Queued", "queued chip card")
+	require.NotContains(t, html, "Summarizing", "not processing")
+	require.NotContains(t, html, "every 2s", "queued card does not poll")
+}
+
+func TestStatusNotFound(t *testing.T) {
+	app := newApp(t)
+	resetDB(t, rawPool(t))
+	rec := getStatus(t, app, videoX)
+	require.Equal(t, http.StatusNotFound, rec.Code)
+}
+
+func getStatus(t *testing.T, app *web.App, videoID string) *httptest.ResponseRecorder {
+	t.Helper()
+	req := httptest.NewRequest(http.MethodGet, "/v/"+videoID+"/status", nil)
+	req.Header.Set("HX-Request", "true")
+	rec := httptest.NewRecorder()
+	app.Router().ServeHTTP(rec, req)
+	return rec
+}

internal/web/processing_test.go

@@ -0,0 +1,27 @@
+package web
+
+import "testing"
+
+func TestProcessingSetAddHasRemove(t *testing.T) {
+	var s ProcessingSet // zero value is usable
+
+	key := processingKey("user-1", "video-1")
+	if s.Has(key) {
+		t.Fatal("fresh set must not report a key as in-flight")
+	}
+
+	s.Add(key)
+	if !s.Has(key) {
+		t.Fatal("Add must mark the key in-flight")
+	}
+
+	// A different user with the same video id is a distinct key.
+	if s.Has(processingKey("user-2", "video-1")) {
+		t.Fatal("keys must be scoped by user")
+	}
+
+	s.Remove(key)
+	if s.Has(key) {
+		t.Fatal("Remove must clear the key")
+	}
+}

internal/web/register_test.go

@@ -7,6 +7,7 @@ import (
 	"strings"
 	"testing"
 
+	"github.com/jackc/pgx/v5/pgxpool"
 	"github.com/stretchr/testify/require"
 )
 
@@ -45,7 +46,7 @@ func TestRegisterCreatesExactlyOneUserAndIdentity(t *testing.T) {
 	truncateAll(t, p)
 
 	req := httptest.NewRequest(http.MethodPost, "/register",
-		strings.NewReader("display_name=Newbie&accept_terms=yes"))
+		strings.NewReader("display_name=Newbie"))
 	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
 	rec := do(t, app, req)
 	require.Equal(t, http.StatusSeeOther, rec.Code)
@@ -74,12 +75,54 @@ func TestRegisterCreatesExactlyOneUserAndIdentity(t *testing.T) {
 	require.Equal(t, 1, totalIdents)
 }
 
+// loginEventCount counts login_events for the stub user via the raw pool.
+func loginEventCount(t *testing.T, p *pgxpool.Pool) int {
+	t.Helper()
+	var n int
+	require.NoError(t, p.QueryRow(context.Background(),
+		`SELECT count(*) FROM login_events WHERE user_id = $1`, userID).Scan(&n))
+	return n
+}
+
+// TestGateStampsLoginEventThrottled: a gated request for a registered user stamps
+// exactly one login event, and a same-day repeat is throttled to no new row — the
+// read-side Stage-0 signal flowing from the registration gate.
+func TestGateStampsLoginEventThrottled(t *testing.T) {
+	app := newApp(t) // stubSubject → userID
+	p := rawPool(t)
+	resetDB(t, p)
+	require.Equal(t, 0, loginEventCount(t, p))
+
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Equal(t, 1, loginEventCount(t, p), "a gated request must stamp one login event")
+
+	do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, 1, loginEventCount(t, p), "a same-day repeat must not stamp again")
+}
+
+// TestUnregisteredSubjectIsNotStamped: a subject with no tapir user is redirected
+// to /register and never reaches the stamp (no user_id to attribute it to).
+func TestUnregisteredSubjectIsNotStamped(t *testing.T) {
+	app := newAppAs(t, "unregistered-sub")
+	p := rawPool(t)
+	truncateAll(t, p)
+
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, http.StatusFound, rec.Code)
+
+	var n int
+	require.NoError(t, p.QueryRow(context.Background(),
+		`SELECT count(*) FROM login_events`).Scan(&n))
+	require.Equal(t, 0, n, "an unregistered subject must not stamp a login event")
+}
+
 func TestRegisterRejectsMissingFields(t *testing.T) {
 	app := newAppAs(t, "incomplete-subject")
 	truncateAll(t, rawPool(t))
 
 	req := httptest.NewRequest(http.MethodPost, "/register",
-		strings.NewReader("display_name=&accept_terms=")) // both missing
+		strings.NewReader("display_name=")) // missing display name
 	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
 	rec := do(t, app, req)
 	require.Equal(t, http.StatusBadRequest, rec.Code)

internal/web/registration.go

@@ -64,6 +64,13 @@ func (a *App) registrationGate(h http.Handler) http.Handler {
 			http.Redirect(w, r, registerPath, http.StatusFound)
 			return
 		}
+		// Stamp the read-side usage signal (Stage-0 gate, ADR-016): the store
+		// throttles this to one row per user per day, so a stamp on every gated
+		// request is cheap. Best-effort — a stamp failure must never break the
+		// request the user actually came for, so it is logged and swallowed.
+		if err := a.Store.StampLogin(r.Context(), userID); err != nil {
+			a.logger().Warn("stamp login event", "user", userID, "err", err)
+		}
 		h.ServeHTTP(w, r.WithContext(withUserID(r.Context(), userID)))
 	})
 }
@@ -108,10 +115,9 @@ func (a *App) handleRegister(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 	displayName := strings.TrimSpace(r.FormValue("display_name"))
-	accepted := r.FormValue("accept_terms") != ""
+	if displayName == "" {
-	if displayName == "" || !accepted {
 		a.renderStatus(w, r, http.StatusBadRequest,
-			RegisterPage(user.Email, "Enter a display name and accept the terms to continue."))
+			RegisterPage(user.Email, "Enter a display name to continue."))
 		return
 	}
 

internal/web/videocard_internal_test.go

@@ -0,0 +1,158 @@
+package web
+
+import (
+	"context"
+	"strings"
+	"testing"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+)
+
+func renderVideoCard(t *testing.T, r store.SummaryRow) string {
+	t.Helper()
+	var sb strings.Builder
+	if err := VideoCard(r).Render(context.Background(), &sb); err != nil {
+		t.Fatalf("render VideoCard: %v", err)
+	}
+	return sb.String()
+}
+
+// noActionButton asserts the rendered HTML contains no form POST (no action URL
+// of the form /v/.../summarize or /v/.../retry-now) and no btn-quiet button.
+func noActionButton(t *testing.T, html, state string) {
+	t.Helper()
+	if strings.Contains(html, "/summarize") {
+		t.Errorf("state %q: expected no summarize URL, got:\n%s", state, html)
+	}
+	if strings.Contains(html, "/retry-now") {
+		t.Errorf("state %q: expected no retry-now URL, got:\n%s", state, html)
+	}
+	if strings.Contains(html, "btn-quiet") {
+		t.Errorf("state %q: expected no action button, got:\n%s", state, html)
+	}
+}
+
+// TestVideoCard_State1_Summarized — chip + no nudge button.
+func TestVideoCard_State1_Summarized(t *testing.T) {
+	html := renderVideoCard(t, store.SummaryRow{
+		VideoID:    "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa",
+		Title:      "Done Video",
+		Summarized: true,
+		Summary:    "A great talk about Go.",
+		AIProvider: "local",
+	})
+
+	if !strings.Contains(html, "local") {
+		t.Errorf("state 1: expected AI provider chip, got:\n%s", html)
+	}
+	noActionButton(t, html, "summarized")
+	if strings.Contains(html, "Summarize") {
+		t.Errorf("state 1: no nudge button on a summarized card, got:\n%s", html)
+	}
+}
+
+// TestVideoCard_State2_NoTranscript — terminal; muted status, NO button, NO POST URL.
+func TestVideoCard_State2_NoTranscript(t *testing.T) {
+	html := renderVideoCard(t, store.SummaryRow{
+		VideoID:          "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa",
+		Title:            "Silent Video",
+		Summarized:       false,
+		TranscriptStatus: "none",
+	})
+
+	if !strings.Contains(html, "No transcript available") {
+		t.Errorf("state 2: expected 'No transcript available' text, got:\n%s", html)
+	}
+	noActionButton(t, html, "none-transcript")
+	if strings.Contains(html, "Summarize") {
+		t.Errorf("state 2: no nudge button when there are no captions, got:\n%s", html)
+	}
+}
+
+// TestVideoCard_State3_Queued — "Queued" chip, no button.
+func TestVideoCard_State3_Queued(t *testing.T) {
+	html := renderVideoCard(t, store.SummaryRow{
+		VideoID:            "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa",
+		Title:              "Queued Video",
+		Summarized:         false,
+		SummarizeRequested: true,
+	})
+
+	if !strings.Contains(html, "Queued") {
+		t.Errorf("state 3: expected 'Queued' chip, got:\n%s", html)
+	}
+	noActionButton(t, html, "queued")
+}
+
+// TestVideoCard_State4_RateLimited — quiet status + "Summarize" → retry-now URL.
+func TestVideoCard_State4_RateLimited(t *testing.T) {
+	html := renderVideoCard(t, store.SummaryRow{
+		VideoID:          "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa",
+		Title:            "Throttled Video",
+		Summarized:       false,
+		TranscriptStatus: "rate_limited",
+	})
+
+	if !strings.Contains(html, "In queue") {
+		t.Errorf("state 4: expected 'In queue' status text, got:\n%s", html)
+	}
+	if !strings.Contains(html, "Summarize") {
+		t.Errorf("state 4: expected 'Summarize' button, got:\n%s", html)
+	}
+	if !strings.Contains(html, "retry-now") {
+		t.Errorf("state 4: expected retry-now URL in form action, got:\n%s", html)
+	}
+	if strings.Contains(html, "/summarize\"") {
+		t.Errorf("state 4: rate-limited card must not POST to /summarize, got:\n%s", html)
+	}
+	if strings.Contains(html, "Try now") {
+		t.Errorf("state 4: 'Try now' verb must not appear, got:\n%s", html)
+	}
+}
+
+// TestVideoCard_State5_Pending — quiet status + "Summarize" → summarize URL.
+func TestVideoCard_State5_Pending(t *testing.T) {
+	html := renderVideoCard(t, store.SummaryRow{
+		VideoID:    "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa",
+		Title:      "Fresh Video",
+		Summarized: false,
+	})
+
+	if !strings.Contains(html, "Not summarized") {
+		t.Errorf("state 5: expected 'Not summarized' status text, got:\n%s", html)
+	}
+	if !strings.Contains(html, "Summarize") {
+		t.Errorf("state 5: expected 'Summarize' button, got:\n%s", html)
+	}
+	if !strings.Contains(html, "/summarize") {
+		t.Errorf("state 5: expected summarize URL in form action, got:\n%s", html)
+	}
+	if strings.Contains(html, "retry-now") {
+		t.Errorf("state 5: pending card must not POST to /retry-now, got:\n%s", html)
+	}
+	if strings.Contains(html, "Try now") {
+		t.Errorf("state 5: 'Try now' verb must not appear, got:\n%s", html)
+	}
+}
+
+// TestVideoCard_ForbiddenCopyAbsent — the over-promising / jargon phrases the
+// honesty pass removed must not reappear in any rendered card state (UX review
+// A3/A4/A6): "Try now", "Summarize now", "Fetching soon", "the next run".
+func TestVideoCard_ForbiddenCopyAbsent(t *testing.T) {
+	forbidden := []string{"Try now", "Summarize now", "Fetching soon", "the next run"}
+	cases := []store.SummaryRow{
+		{VideoID: "a", Summarized: true, Summary: "s", AIProvider: "local"},
+		{VideoID: "b", TranscriptStatus: "none"},
+		{VideoID: "c", SummarizeRequested: true},
+		{VideoID: "d", TranscriptStatus: "rate_limited"},
+		{VideoID: "e"},
+	}
+	for _, r := range cases {
+		html := renderVideoCard(t, r)
+		for _, phrase := range forbidden {
+			if strings.Contains(html, phrase) {
+				t.Errorf("forbidden copy %q reappeared in card %q:\n%s", phrase, r.VideoID, html)
+			}
+		}
+	}
+}

internal/web/view.go

@@ -4,6 +4,7 @@ import (
 	"regexp"
 	"strings"
 	"time"
+	"unicode/utf8"
 
 	"github.com/a-h/templ"
 
@@ -171,6 +172,151 @@ func actionURL(videoID string) templ.SafeURL {
 	return templ.SafeURL("/v/" + videoID + "/action")
 }
 
+// summarizeURL builds the manual-queue POST path for a video id.
+func summarizeURL(videoID string) templ.SafeURL {
+	return templ.SafeURL("/v/" + videoID + "/summarize")
+}
+
+// statusURL builds the processing-status poll path (GET) for a video id — the
+// HTMX poll target while an immediate summarization is in flight.
+func statusURL(videoID string) templ.SafeURL {
+	return templ.SafeURL("/v/" + videoID + "/status")
+}
+
+// Charmbracelet-inspired palette for the summarizing animation (TapirSpinner) —
+// a charm purple box, pink tapir, mint snout/eyes/progress. Kept as named consts
+// so the inline span colours and the CSS track/fill share one source of truth.
+const (
+	CharmPurple = "#7653FC" // box border
+	CharmPink   = "#FF6E9C" // tapir body
+	CharmMint   = "#0EF9B6" // snout, eyes, progress fill
+	CharmCream  = "#FFFDF5" // bright text
+	CharmDim    = "#6C6C6C" // dim text
+	charmTrack  = "#2D2D2D" // empty progress track (internal: dark char colour)
+)
+
+// tapirInteriorW is the fixed inner width of the Charm box, in monospace cells.
+const tapirInteriorW = 34
+
+// tapirBarFill is the mint progress fill (27 cells), revealed left→right by the
+// CSS width/clip animation over the dim track drawn in each frame.
+const tapirBarFill = "███████████████████████████"
+
+// tapirRun is one coloured (or uncoloured) text segment of a box row.
+type tapirRun struct {
+	s     string
+	color string // "" = no span (plain text)
+}
+
+func tapirSpan(color, s string) string {
+	if color == "" {
+		return s
+	}
+	return `<span style="color:` + color + `">` + s + `</span>`
+}
+
+// tapirLine renders one interior box row: concatenate the coloured runs, pad with
+// spaces to the fixed interior width, then flank with the purple side borders.
+// Padding is computed from the runs' rune counts, so every row's right border
+// lines up no matter how many runs it has (assuming 1-cell monospace glyphs).
+func tapirLine(runs ...tapirRun) string {
+	var b strings.Builder
+	width := 0
+	for _, r := range runs {
+		b.WriteString(tapirSpan(r.color, r.s))
+		width += utf8.RuneCountInString(r.s)
+	}
+	if width < tapirInteriorW {
+		b.WriteString(strings.Repeat(" ", tapirInteriorW-width))
+	}
+	bar := tapirSpan(CharmPurple, "│")
+	return bar + b.String() + bar
+}
+
+// tapirFrameHTML builds one animation frame: a rounded Charm box around a colored
+// ASCII tapir, a dim progress track, and labels. snout is the wiggling nose glyph
+// that differs between the three frames. Returned as raw HTML (coloured spans),
+// emitted verbatim by the template via templ.Raw.
+func tapirFrameHTML(snout string) string {
+	top := tapirSpan(CharmPurple, "╭"+strings.Repeat("─", tapirInteriorW)+"╮")
+	bottom := tapirSpan(CharmPurple, "╰"+strings.Repeat("─", tapirInteriorW)+"╯")
+	lines := []string{
+		top,
+		tapirLine(tapirRun{s: "  "}, tapirRun{s: "◆", color: CharmMint}, tapirRun{s: " "}, tapirRun{s: "tapir", color: CharmCream}),
+		tapirLine(),
+		tapirLine(tapirRun{s: "       "}, tapirRun{s: "▄▄▄▄▄", color: CharmPink}),
+		tapirLine(tapirRun{s: "     "}, tapirRun{s: "▄█▓▓▓▓█▄", color: CharmPink}, tapirRun{s: "   "}, tapirRun{s: snout, color: CharmMint}),
+		tapirLine(tapirRun{s: "    "}, tapirRun{s: "█▓(", color: CharmPink}, tapirRun{s: " "}, tapirRun{s: "◕ ◕", color: CharmMint}, tapirRun{s: ")▓█", color: CharmPink}, tapirRun{s: "──┘", color: CharmMint}, tapirRun{s: "  "}, tapirRun{s: "< thinking...", color: CharmDim}),
+		tapirLine(tapirRun{s: "     "}, tapirRun{s: "▀█▓▓▓▓█▀", color: CharmPink}),
+		tapirLine(tapirRun{s: "      "}, tapirRun{s: "██▄▄██", color: CharmPink}),
+		tapirLine(tapirRun{s: "     "}, tapirRun{s: "▀▀", color: CharmPink}, tapirRun{s: "  "}, tapirRun{s: "▀▀", color: CharmPink}),
+		tapirLine(),
+		tapirLine(tapirRun{s: "  "}, tapirRun{s: "[", color: CharmDim}, tapirRun{s: strings.Repeat("░", 27), color: charmTrack}, tapirRun{s: "]", color: CharmDim}),
+		tapirLine(tapirRun{s: "   "}, tapirRun{s: "summarizing", color: CharmDim}),
+		bottom,
+	}
+	return strings.Join(lines, "\n")
+}
+
+// The three frames differ only in the snout glyph (∩ → ∪ → ~), cross-faded by CSS
+// to read as a tapir wiggling its nose while it thinks.
+var (
+	tapirFrameHTML1 = tapirFrameHTML("∩")
+	tapirFrameHTML2 = tapirFrameHTML("∪")
+	tapirFrameHTML3 = tapirFrameHTML("~")
+)
+
+// welcomeHeroHTML is the static Charm-box tapir mascot on the public landing
+// page — the same rounded purple box / pink tapir / mint accents as the spinner,
+// but a single still frame with a friendly tagline instead of the animation.
+// Built from the shared tapirLine helpers so the aesthetic stays in one place.
+func welcomeHeroHTML() string {
+	top := tapirSpan(CharmPurple, "╭"+strings.Repeat("─", tapirInteriorW)+"╮")
+	bottom := tapirSpan(CharmPurple, "╰"+strings.Repeat("─", tapirInteriorW)+"╯")
+	lines := []string{
+		top,
+		tapirLine(tapirRun{s: "  "}, tapirRun{s: "◆", color: CharmMint}, tapirRun{s: " "}, tapirRun{s: "tapir", color: CharmCream}),
+		tapirLine(),
+		tapirLine(tapirRun{s: "       "}, tapirRun{s: "▄▄▄▄▄", color: CharmPink}),
+		tapirLine(tapirRun{s: "     "}, tapirRun{s: "▄█▓▓▓▓█▄", color: CharmPink}, tapirRun{s: "   "}, tapirRun{s: "∩", color: CharmMint}),
+		tapirLine(tapirRun{s: "    "}, tapirRun{s: "█▓(", color: CharmPink}, tapirRun{s: " "}, tapirRun{s: "◕ ◕", color: CharmMint}, tapirRun{s: ")▓█", color: CharmPink}, tapirRun{s: "──┘", color: CharmMint}),
+		tapirLine(tapirRun{s: "     "}, tapirRun{s: "▀█▓▓▓▓█▀", color: CharmPink}),
+		tapirLine(tapirRun{s: "      "}, tapirRun{s: "██▄▄██", color: CharmPink}),
+		tapirLine(tapirRun{s: "     "}, tapirRun{s: "▀▀", color: CharmPink}, tapirRun{s: "  "}, tapirRun{s: "▀▀", color: CharmPink}),
+		tapirLine(),
+		tapirLine(tapirRun{s: "  "}, tapirRun{s: "watch less, know more", color: CharmMint}),
+		bottom,
+	}
+	return strings.Join(lines, "\n")
+}
+
+var welcomeHero = welcomeHeroHTML()
+
+// summarizeModeLabel names the current mode for display.
+func summarizeModeLabel(auto bool) string {
+	if auto {
+		return "Automatic"
+	}
+	return "Manual"
+}
+
+// summarizeModeToggleLabel is the caption on the toggle button — it names the mode
+// the click switches TO (the opposite of the current one).
+func summarizeModeToggleLabel(auto bool) string {
+	if auto {
+		return "Switch to manual"
+	}
+	return "Switch to automatic"
+}
+
+// boolStr renders a bool as the "enabled" form value the toggle submits.
+func boolStr(b bool) string {
+	if b {
+		return "true"
+	}
+	return "false"
+}
+
 // externalURL passes a stored source URL through templ's URL sanitiser.
 func externalURL(u string) templ.SafeURL {
 	return templ.URL(u)
@@ -236,6 +382,89 @@ func disconnectURL(provider string) templ.SafeURL {
 	return templ.SafeURL("/account/disconnect/" + provider)
 }
 
+// PipelineStats summarises the user's video backlog so the list page can show
+// a one-line status bar ("2 summaries · 256 fetching soon · 12 no captions").
+type PipelineStats struct {
+	Summarized  int
+	RateLimited int // in the backoff window, will be retried
+	NoText      int // no caption track available
+	Pending     int // discovered but not yet attempted
+}
+
+// pipelineStats computes a PipelineStats from all (unfiltered) rows.
+func pipelineStats(rows []store.SummaryRow) PipelineStats {
+	var s PipelineStats
+	for _, r := range rows {
+		switch {
+		case r.Summarized:
+			s.Summarized++
+		case r.TranscriptStatus == "rate_limited":
+			s.RateLimited++
+		case r.TranscriptStatus == "none":
+			s.NoText++
+		default:
+			s.Pending++
+		}
+	}
+	return s
+}
+
+// retryNowURL builds the POST path for manual retry of a rate-limited video.
+func retryNowURL(videoID string) templ.SafeURL {
+	return templ.SafeURL("/v/" + videoID + "/retry-now")
+}
+
+// listBuckets splits the (already filtered) video list into what the list view
+// shows where, so the readable summaries are not buried under the un-summarized
+// back-catalogue (UX review B3/B4). It is one feed with a noise-collapse, not
+// separate sections:
+//   - Main: summarized videos + recent un-summarized ones — shown inline as cards.
+//   - Older: un-summarized videos published before the recency cutoff — collapsed
+//     behind a single "Show N older videos" disclosure (they will not auto-fill;
+//     they are summarize-on-demand).
+//   - NoCaption: count of un-summarized videos with no caption track — collapsed
+//     to one honest line instead of N dead terminal cards.
+type listBuckets struct {
+	Main      []store.SummaryRow
+	Older     []store.SummaryRow
+	NoCaption int
+}
+
+// bucketRows classifies rows into the list buckets given a recency cutoff. A zero
+// cutoff (recency collapse disabled) leaves Older empty — every un-summarized,
+// captioned video stays inline. Order within each bucket is preserved.
+func bucketRows(rows []store.SummaryRow, cutoff time.Time) listBuckets {
+	var b listBuckets
+	for _, r := range rows {
+		switch {
+		case r.Summarized:
+			b.Main = append(b.Main, r)
+		case r.TranscriptStatus == "none":
+			b.NoCaption++
+		case isOlder(r, cutoff):
+			b.Older = append(b.Older, r)
+		default:
+			b.Main = append(b.Main, r)
+		}
+	}
+	return b
+}
+
+// isOlder reports whether an un-summarized row falls before the recency cutoff.
+// A zero cutoff (window disabled) or an undated row is never "older" — it cannot
+// be aged out, so it stays inline rather than being hidden in the disclosure.
+func isOlder(r store.SummaryRow, cutoff time.Time) bool {
+	if cutoff.IsZero() || r.PublishedAt.IsZero() {
+		return false
+	}
+	return r.PublishedAt.Before(cutoff)
+}
+
+// empty reports whether there is nothing to show at all (drives the empty state).
+func (b listBuckets) empty() bool {
+	return len(b.Main) == 0 && len(b.Older) == 0 && b.NoCaption == 0
+}
+
 // Filter holds the list-view query parameters. Empty fields mean "no constraint".
 // Dates are kept as the raw YYYY-MM-DD strings so the form re-renders the user's
 // input verbatim; parsing happens in matchFilter.
@@ -243,6 +472,15 @@ type Filter struct {
 	Channel        string
 	From           string
 	To             string
+	OnlySummarized bool // show only videos that have a summary
+}
+
+// active reports whether any filter constraint is set. Drives whether the filter
+// bar is shown at all: on a genuinely empty account (no rows AND no active
+// filter) the bar is hidden so the connect CTA stands alone (UX review C1); a
+// filter that happens to match nothing still shows the bar so it can be cleared.
+func (f Filter) active() bool {
+	return f.Channel != "" || f.From != "" || f.To != "" || f.OnlySummarized
 }
 
 // matches reports whether a row satisfies the filter. Channel is an exact match;
@@ -250,6 +488,9 @@ type Filter struct {
 // (no constraint) — Stage-0 filtering is in-memory over the listed rows, not a
 // store query.
 func (f Filter) matches(r store.SummaryRow) bool {
+	if f.OnlySummarized && !r.Summarized {
+		return false
+	}
 	if f.Channel != "" && r.Channel != f.Channel {
 		return false
 	}
@@ -280,7 +521,7 @@ func parseDate(s string) (time.Time, bool) {
 
 // apply returns the subset of rows matching the filter, preserving order.
 func (f Filter) apply(rows []store.SummaryRow) []store.SummaryRow {
-	if f == (Filter{}) {
+	if f.Channel == "" && f.From == "" && f.To == "" && !f.OnlySummarized {
 		return rows
 	}
 	out := rows[:0:0]
@@ -332,7 +573,13 @@ main { max-width: 60rem; margin: 0 auto; padding: var(--s4) var(--s3); }
 .filters label { display: flex; flex-direction: column; font-size: .78rem; text-transform: uppercase; letter-spacing: .04em; color: var(--muted); gap: var(--s1); }
 .filters input { font: inherit; padding: .4rem .55rem; border: 1px solid var(--line); border-radius: var(--radius); background: var(--card); color: var(--fg); min-width: 9rem; }
 .filters input:focus-visible { outline: 2px solid var(--accent); outline-offset: 1px; border-color: var(--accent); }
+.filter-check { flex-direction: row !important; align-items: center; gap: var(--s2) !important; padding-bottom: .45rem; }
+.filter-check input[type=checkbox] { width: 1rem; height: 1rem; min-width: 0; padding: 0; accent-color: var(--accent); cursor: pointer; }
 .btn { font: inherit; font-weight: 600; padding: .45rem 1rem; border: 1px solid var(--accent); border-radius: var(--radius); background: var(--accent); color: var(--accent-fg); cursor: pointer; }
+/* anchors styled as buttons: the generic a{} / a:visited{} colour rules outrank
+   .btn on <a>, painting the label accent-on-accent (invisible). Restore the
+   button foreground for anchor buttons, visited included. */
+a.btn, a.btn:visited { color: var(--accent-fg); }
 .btn:hover { filter: brightness(1.05); }
 .btn:active { transform: translateY(1px); }
 
@@ -344,13 +591,72 @@ main { max-width: 60rem; margin: 0 auto; padding: var(--s4) var(--s3); }
 .card-preview { color: var(--muted); font-size: .9rem; line-height: 1.5; display: -webkit-box; -webkit-line-clamp: 1; line-clamp: 1; -webkit-box-orient: vertical; overflow: hidden; }
 .card-foot { display: flex; gap: var(--s2); align-items: center; flex-wrap: wrap; margin-top: var(--s1); }
 .chip { display: inline-block; padding: .15rem .55rem; border-radius: 999px; background: var(--accent-weak); color: var(--accent); font-size: .72rem; font-weight: 600; }
+/* passive "retrying later" chip: dim/grey (CharmDim), not the accent — it is a
+   status, not an action the user can take. */
+.chip-retry { background: rgba(108, 108, 108, .16); color: #6c6c6c; }
+.pipeline-bar { display: flex; gap: var(--s3); align-items: center; flex-wrap: wrap; margin-bottom: var(--s3); font-size: .8rem; color: var(--muted); }
+.pipeline-bar span { display: flex; align-items: center; gap: var(--s1); }
+.pipeline-bar span + span::before { content: "·"; margin-right: var(--s1); }
+.pipeline-note { margin: calc(-1 * var(--s2)) 0 var(--s3); font-size: .8rem; line-height: 1.5; max-width: 40rem; }
+/* one-line count of caption-less videos (collapsed instead of N dead cards) */
+.list-note { margin: var(--s3) 0 0; font-size: .85rem; }
+/* older un-summarized back-catalogue, collapsed behind a disclosure so it does
+   not bury the readable summaries above it */
+.older-videos { margin-top: var(--s4); }
+.older-videos > summary { cursor: pointer; font-size: .85rem; font-weight: 600; color: var(--accent); padding: var(--s2) 0; list-style: revert; }
+.older-videos > summary:hover { text-decoration: underline; }
+.older-videos[open] > summary { margin-bottom: var(--s3); }
+.older-videos .cards { margin-top: 0; }
+.card-nudge-form { display: inline; }
+.btn-quiet { font: inherit; font-size: .72rem; font-weight: 600; padding: .15rem .55rem; border-radius: 999px; border: 1px solid var(--accent); background: transparent; color: var(--accent); cursor: pointer; }
+.btn-quiet:hover { background: var(--accent-weak); }
+.chip-warn { background: rgba(255, 110, 156, .15); color: #FF6E9C; }
 .card-state { color: var(--muted); font-size: .8rem; }
 .badge { display: inline-block; padding: .15rem .55rem; border-radius: 999px; background: var(--badge-bg); color: var(--badge-fg); font-size: .72rem; font-weight: 600; }
 
+/* pending (discovered-but-unsummarized) card: muted until summarized */
+.card-pending { border-style: dashed; }
+.card-pending .card-title { color: var(--muted); font-weight: 600; }
+
+/* summarizing animation — a Charmbracelet-style TUI panel rendered in the
+   browser: a dark terminal card, a rounded purple box around a pink ASCII tapir,
+   and a lipgloss-style progress bar. Three frames are stacked and cross-faded by
+   a stepped keyframe (staggered delays) so the snout appears to wiggle; the
+   progress fill grows independently via a clip animation over the dim track. */
+.card-processing { border-style: dashed; }
+.tapir-charm { position: relative; display: inline-block; background: #0d0d12; border-radius: 10px; padding: .8em 1em; margin: var(--s2) 0; font: .82rem/1.15 ui-monospace, SFMono-Regular, Menlo, "Cascadia Code", monospace; box-shadow: 0 2px 14px rgba(118, 83, 252, .25); }
+.tapir-charm pre { margin: 0; white-space: pre; opacity: 0; animation: tapir-cycle 1.2s steps(1, end) infinite; }
+.tapir-charm .tapir-f1 { position: relative; animation-delay: 0s; }
+.tapir-charm .tapir-f2 { position: absolute; top: .8em; left: 1em; animation-delay: .4s; }
+.tapir-charm .tapir-f3 { position: absolute; top: .8em; left: 1em; animation-delay: .8s; }
+@keyframes tapir-cycle { 0%, 33.32% { opacity: 1; } 33.33%, 100% { opacity: 0; } }
+/* progress fill: 27 mint cells overlaying the dim track at box row 10, col 3,
+   revealed left→right over 8s, looping. */
+.tapir-bar { position: absolute; top: calc(.8em + 11.5em); left: calc(1em + 3ch); height: 1.15em; line-height: 1.15; overflow: hidden; }
+.tapir-bar-fill { animation: tapir-fill 8s linear infinite; text-shadow: 0 0 6px rgba(14, 249, 182, .7); }
+@keyframes tapir-fill { 0% { clip-path: inset(0 100% 0 0); } 100% { clip-path: inset(0 0 0 0); } }
+.tapir-label { color: var(--muted); font-size: .9rem; margin: 0; }
+@media (prefers-reduced-motion: reduce) {
+  .tapir-charm pre { animation: none; }
+  .tapir-charm .tapir-f2, .tapir-charm .tapir-f3 { display: none; }
+  .tapir-charm .tapir-f1 { opacity: 1; }
+  .tapir-bar-fill { animation: none; clip-path: inset(0 35% 0 0); }
+}
+
+/* summarization mode toggle on the account page */
+.summarize-mode { display: flex; gap: var(--s3); align-items: center; flex-wrap: wrap; }
+.summarize-mode p { margin: 0; }
+.summarize-mode form { margin: 0; }
+
 /* empty state */
 .empty { text-align: center; color: var(--muted); padding: var(--s5) var(--s4); border: 1px dashed var(--line); border-radius: var(--radius); background: var(--card); }
 .empty strong { display: block; color: var(--fg); font-size: 1.05rem; margin-bottom: var(--s2); }
 .empty code { background: var(--accent-weak); color: var(--accent); padding: .1rem .35rem; border-radius: .3rem; }
+.empty p { margin: var(--s3) 0 0; }
+/* connected-but-empty: a distinct accent callout, not a muted blank state, so a
+   fresh account knows the next step is to run tapir, not "something is broken". */
+.empty-connected { border-style: solid; border-color: var(--accent); background: var(--accent-weak); color: var(--fg); }
+.empty-connected strong { color: var(--accent); }
 
 /* flash / notification banner */
 .flash { padding: var(--s2) var(--s3); border-radius: var(--radius); margin-bottom: var(--s4); font-size: .92rem; border: 1px solid var(--line); }
@@ -366,6 +672,7 @@ main { max-width: 60rem; margin: 0 auto; padding: var(--s4) var(--s3); }
 
 /* detail reader */
 .detail { max-width: 38rem; }
+.detail .back { margin: 0 0 var(--s3); font-size: .85rem; }
 .detail h1 { font-size: 1.7rem; line-height: 1.25; margin: 0 0 var(--s2); }
 .detail .meta { color: var(--muted); font-size: .9rem; margin: 0 0 var(--s2); display: flex; gap: var(--s2); align-items: center; flex-wrap: wrap; }
 .detail .source { margin: 0 0 var(--s4); font-size: .9rem; }
@@ -377,8 +684,13 @@ main { max-width: 60rem; margin: 0 auto; padding: var(--s4) var(--s3); }
 .detail ul { margin: 0; padding-left: 1.2rem; line-height: 1.6; }
 .detail li { margin-bottom: var(--s1); }
 
-/* action toggles */
+/* action toggles — watched|skipped form one segmented control (they are mutually
-.actions { display: flex; gap: var(--s2); margin: var(--s4) 0; flex-wrap: wrap; }
+   exclusive), "saved" sits apart as an independent toggle */
+.actions { display: flex; gap: var(--s3); margin: var(--s4) 0; flex-wrap: wrap; align-items: center; }
+.segmented { display: inline-flex; }
+.segmented .action { border-radius: 0; border-right-width: 0; }
+.segmented .action:first-child { border-top-left-radius: var(--radius); border-bottom-left-radius: var(--radius); }
+.segmented .action:last-child { border-top-right-radius: var(--radius); border-bottom-right-radius: var(--radius); border-right-width: 1px; }
 .actions .action { font: inherit; padding: .4rem .9rem; border: 1px solid var(--line); border-radius: var(--radius); background: var(--card); color: var(--fg); cursor: pointer; transition: border-color .15s, background .15s; }
 .actions .action:hover { border-color: var(--accent); }
 .actions .action:focus-visible { outline: 2px solid var(--accent); outline-offset: 1px; }
@@ -393,6 +705,11 @@ main { max-width: 60rem; margin: 0 auto; padding: var(--s4) var(--s3); }
 .account-meta { display: grid; grid-template-columns: max-content 1fr; gap: var(--s1) var(--s3); margin: 0; }
 .account-meta dt { color: var(--muted); font-size: .85rem; }
 .account-meta dd { margin: 0; }
+.channel-errors { }
+.channel-error-list { list-style: none; margin: 0 0 var(--s3); padding: 0; display: grid; gap: var(--s1); }
+.channel-error-list li { display: flex; align-items: center; gap: var(--s2); }
+.channel-error-name { font-weight: 500; }
+.channel-error-since { font-size: .8rem; }
 .conn-list { list-style: none; margin: 0 0 var(--s3); padding: 0; display: grid; gap: var(--s2); }
 .conn { background: var(--card); border: 1px solid var(--line); border-radius: var(--radius); padding: var(--s3); display: flex; flex-direction: column; gap: var(--s1); }
 .conn-main { display: flex; gap: var(--s2); align-items: center; flex-wrap: wrap; }
@@ -419,6 +736,16 @@ main { max-width: 60rem; margin: 0 auto; padding: var(--s4) var(--s3); }
   .confirm-delete > summary:hover { background: #3a1714; }
 }
 
+/* public landing page (/welcome) — the Charm-box mascot hero plus the sign-in CTA */
+.welcome { text-align: center; padding: var(--s5) var(--s3); display: flex; flex-direction: column; align-items: center; gap: var(--s4); }
+.welcome-hero { background: #0d0d12; border-radius: 10px; padding: .9em 1.1em; display: inline-block; box-shadow: 0 2px 14px rgba(118, 83, 252, .25); }
+.welcome-hero pre { margin: 0; white-space: pre; font: .82rem/1.15 ui-monospace, SFMono-Regular, Menlo, "Cascadia Code", monospace; }
+.welcome-title { font-size: 1.9rem; line-height: 1.2; margin: 0; }
+.welcome-tagline { color: var(--muted); font-size: 1.05rem; line-height: 1.5; margin: 0; max-width: 32rem; }
+.welcome-cta { display: flex; gap: var(--s3); flex-wrap: wrap; justify-content: center; align-items: center; }
+.welcome-sub { color: var(--muted); font-size: .9rem; margin: 0; }
+.btn-lg { padding: .6rem 1.6rem; font-size: 1.05rem; }
+
 @media (max-width: 640px) {
   main { padding: var(--s3) var(--s2); }
   .filters { gap: var(--s2); }

internal/web/view_internal_test.go

@@ -1,6 +1,11 @@
 package web
 
-import "testing"
+import (
+	"regexp"
+	"strings"
+	"testing"
+	"unicode/utf8"
+)
 
 func TestEmbedURL(t *testing.T) {
 	tests := []struct {
@@ -27,3 +32,19 @@ func TestEmbedURL(t *testing.T) {
 		})
 	}
 }
+
+// TestTapirFrameRowsAligned asserts every box row has the same cell width once
+// the inline-colour spans are stripped, so the rounded border lines up on every
+// line (the panel only looks right if the right │ is flush across all rows).
+func TestTapirFrameRowsAligned(t *testing.T) {
+	stripSpan := regexp.MustCompile(`</?span[^>]*>`)
+	for name, frame := range map[string]string{"f1": tapirFrameHTML1, "f2": tapirFrameHTML2, "f3": tapirFrameHTML3} {
+		plain := stripSpan.ReplaceAllString(frame, "")
+		want := tapirInteriorW + 2 // both purple side borders
+		for i, line := range strings.Split(plain, "\n") {
+			if got := utf8.RuneCountInString(line); got != want {
+				t.Errorf("%s line %d width = %d, want %d: %q", name, i, got, want, line)
+			}
+		}
+	}
+}

internal/web/views.templ

@@ -1,6 +1,7 @@
 package web
 
 import (
+	"fmt"
 	"strings"
 
 	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
@@ -22,7 +23,7 @@ templ Layout(title string) {
 		<body>
 			<header>
 				<a href="/" class="brand">Tapir</a>
-				<nav class="nav"><a href="/account">Account</a></nav>
+				<nav class="nav"><a href="/account">Account</a><a href="/auth/logout">Log out</a></nav>
 			</header>
 			<main>
 				{ children... }
@@ -31,6 +32,64 @@ templ Layout(title string) {
 	</html>
 }
 
+// PublicLayout is the shell for unauthenticated pages (/welcome, /invite).
+// Same structure as Layout but without the nav auth links — a visitor who is not
+// logged in should not see "Account" or "Log out".
+templ PublicLayout(title string) {
+	<!DOCTYPE html>
+	<html lang="en">
+		<head>
+			<meta charset="utf-8"/>
+			<meta name="viewport" content="width=device-width, initial-scale=1"/>
+			<title>{ title }</title>
+			<script src="/static/htmx.min.js" defer></script>
+			@templ.Raw(styleTag)
+		</head>
+		<body>
+			<header>
+				<a href="/" class="brand">Tapir</a>
+			</header>
+			<main>
+				{ children... }
+			</main>
+		</body>
+	</html>
+}
+
+// WelcomePage is the public landing page (served at /welcome, outside the auth
+// guard — ADR-012). Logged out: the tapir mascot, a one-line tagline, and a
+// single "Get Started" CTA into the shared Dex flow (sign-in and sign-up are the
+// same URL). Logged in: a greeting plus links back into the app and to log out.
+templ WelcomePage(user User, loggedIn bool) {
+	@PublicLayout("Tapir — Watch less, know more") {
+		<section class="welcome">
+			<div class="welcome-hero">
+				<pre aria-hidden="true">@templ.Raw(welcomeHero)</pre>
+			</div>
+			if loggedIn {
+				<h1 class="welcome-title">Welcome back</h1>
+				if user.Email != "" {
+					<p class="welcome-tagline">Signed in as { user.Email }.</p>
+				}
+				<div class="welcome-cta">
+					<a class="btn btn-lg" href="/">Go to my Tapir</a>
+					<a class="btn-secondary" href="/auth/logout">Log Out</a>
+				</div>
+			} else {
+				<h1 class="welcome-title">Watch less, know more</h1>
+				<p class="welcome-tagline">
+					Tapir summarizes the videos your subscriptions publish, so you can
+					skim the gist and decide what is worth your time.
+				</p>
+				<div class="welcome-cta">
+					<a class="btn btn-lg" href="/auth/login">Get Started</a>
+				</div>
+				<p class="welcome-sub">Tapir is invite-only right now. If you've been invited, sign in above. New summaries land gradually — Tapir fetches captions slowly to respect YouTube's limits.</p>
+			}
+		</section>
+	}
+}
+
 // flashBanner renders a one-shot notification for a flash code (connect success/
 // failure, disconnect, delete, registration). An empty or unknown code renders
 // nothing, so it is safe to drop into any page unconditionally. Reused across the
@@ -45,16 +104,45 @@ templ flashBanner(code string) {
 // #summary-list region; a non-HTMX request renders the whole page. flash carries
 // a one-shot notification (e.g. "connected", "registered") surfaced on arrival
 // after a POST→redirect.
-templ ListPage(rows []store.SummaryRow, f Filter, flash string) {
+templ ListPage(b listBuckets, f Filter, stats PipelineStats, flash string, hasConnected bool) {
 	@Layout("Tapir — Summaries") {
 		@flashBanner(flash)
+		if !b.empty() || f.active() {
 			@filterForm(f)
+		}
+		if stats.RateLimited > 0 || stats.Pending > 0 || stats.NoText > 0 {
+			@pipelineBar(stats)
+		}
+		if stats.RateLimited+stats.Pending > 0 {
+			<p class="pipeline-note muted">
+				Tapir fetches captions slowly on purpose, to respect YouTube's limits —
+				new summaries land gradually. Check back tomorrow.
+			</p>
+		}
 		<div id="summary-list">
-			@summaryList(rows)
+			@summaryList(b, hasConnected)
 		</div>
 	}
 }
 
+// pipelineBar is the one-line backlog status. Counts are framed by what the user
+// can read NOW ("ready"), what is waiting behind the honest caption rate limit
+// ("in queue" = pending + rate-limited, never "fetching soon" — see A3/ADR-014),
+// and what is permanently unreadable ("no captions").
+templ pipelineBar(s PipelineStats) {
+	<div class="pipeline-bar">
+		if s.Summarized > 0 {
+			<span>{ fmt.Sprintf("%d ready", s.Summarized) }</span>
+		}
+		if s.RateLimited+s.Pending > 0 {
+			<span>{ fmt.Sprintf("%d in queue", s.RateLimited+s.Pending) }</span>
+		}
+		if s.NoText > 0 {
+			<span class="muted">{ fmt.Sprintf("%d no captions", s.NoText) }</span>
+		}
+	</div>
+}
+
 templ filterForm(f Filter) {
 	<form
 		class="filters"
@@ -66,35 +154,84 @@ templ filterForm(f Filter) {
 		hx-indicator="#filter-indicator"
 	>
 		<label>Channel <input type="text" name="channel" value={ f.Channel } placeholder="any"/></label>
-		<label>From <input type="date" name="from" value={ f.From }/></label>
+		<label class="filter-check">
-		<label>To <input type="date" name="to" value={ f.To }/></label>
+			<input type="checkbox" name="summarized" value="1" if f.OnlySummarized { checked }/>
+			Summarized only
+		</label>
 		<button type="submit" class="btn">Filter</button>
 		<span id="filter-indicator" class="htmx-indicator">filtering…</span>
 	</form>
 }
 
-// summaryList is the swappable list fragment: one card per summary (title link,
+// summaryList is the swappable list fragment. It leads with readable summaries +
-// channel · date meta, provider chip, fallback badge, action state). Cards
+// recent un-summarized cards (b.Main), then collapses the noise so it does not
+// bury the payload (UX review B3/B4): a one-line count of caption-less videos,
+// and a single disclosure holding the older un-summarized back-catalogue. Cards
 // reflow to a single column on mobile; an empty list shows a friendly first-run
 // state instead of a blank table.
-templ summaryList(rows []store.SummaryRow) {
+templ summaryList(b listBuckets, hasConnected bool) {
-	if len(rows) == 0 {
+	if b.empty() {
-		<div class="empty">
+		if hasConnected {
-			<strong>No summaries yet</strong>
+			<div class="empty empty-connected">
-			<span>Summaries appear here as your subscriptions are processed — run <code>tapir run</code> to fetch and summarize new videos.</span>
+				<strong>Your account is connected</strong>
+				<span>Tapir is finding your subscriptions and fetching captions — summaries appear here gradually. Check back later.</span>
 			</div>
+		} else {
+			<div class="empty">
+				<strong>No videos yet</strong>
+				<span>Connect your YouTube account to get started.</span>
+				<p><a class="btn" href="/oauth/youtube/connect">Connect YouTube</a></p>
+			</div>
+		}
 	} else {
 		<ul class="cards">
-			for _, r := range rows {
+			for _, r := range b.Main {
-				<li class="card">
+				@VideoCard(r)
+			}
+		</ul>
+		if b.NoCaption > 0 {
+			<p class="list-note muted">{ fmt.Sprintf("%d video(s) have no captions and can't be summarized.", b.NoCaption) }</p>
+		}
+		if len(b.Older) > 0 {
+			<details class="older-videos">
+				<summary>{ fmt.Sprintf("Show %d older videos — summarize on demand", len(b.Older)) }</summary>
+				<ul class="cards">
+					for _, r := range b.Older {
+						@VideoCard(r)
+					}
+				</ul>
+			</details>
+		}
+	}
+}
+
+// VideoCard is one list card, returned standalone by POST /v/{id}/summarize and
+// /v/{id}/retry-now (HTMX swaps outerHTML). Five footer states, status-primary:
+//  1. Summarized — preview + chip + actions; no button.
+//  2. No captions (TranscriptStatus=="none") — terminal; "No transcript available"; no button.
+//  3. Queued (SummarizeRequested) — "Queued · summarizing shortly"; no button.
+//  4. Rate-limited — "In queue" + quiet "Summarize" → /retry-now.
+//  5. Pending (else) — "Not summarized" + quiet "Summarize" → /summarize.
+// States 4 and 5 use one verb ("Summarize") and one style (.btn-quiet); the
+// backend side-effect difference (clear-backoff vs. set-flag) is invisible to users.
+templ VideoCard(r store.SummaryRow) {
+	<li class={ "card", templ.KV("card-pending", !r.Summarized) } id={ "video-" + r.VideoID }>
+		if r.Summarized {
 			<div class="card-title"><a href={ videoURL(r.VideoID) }>{ displayTitle(r) }</a></div>
+		} else {
+			<div class="card-title">{ displayTitle(r) }</div>
+		}
 		if cardMeta(r) != "" {
 			<div class="card-meta">{ cardMeta(r) }</div>
 		}
+		if r.Summarized {
 			if p := previewText(r.Summary, 160); p != "" {
 				<div class="card-preview">{ p }</div>
 			}
+		}
 		<div class="card-foot">
+			if r.Summarized {
+				// State 1: summarized — provider chip, fallback badge, action state.
 				if r.AIProvider != "" {
 					<span class="chip">{ r.AIProvider }</span>
 				}
@@ -104,11 +241,79 @@ templ summaryList(rows []store.SummaryRow) {
 				if len(r.Actions) > 0 {
 					<span class="card-state">{ strings.Join(r.Actions, ", ") }</span>
 				}
+			} else if r.TranscriptStatus == "none" {
+				// State 2: no captions — terminal dead-end; nothing the user can do.
+				<span class="card-state muted">No transcript available</span>
+			} else if r.SummarizeRequested {
+				// State 3: queued — being summarized on the next pass; no scheduler jargon.
+				<span class="chip">Queued</span>
+				<span class="card-state muted">summarizing shortly</span>
+			} else if r.TranscriptStatus == "rate_limited" {
+				// State 4: rate-limited — honest "in queue" status (NOT "fetching soon",
+				// which oversells imminence) + a quiet nudge → retry-now handler.
+				<span class="card-state muted">In queue</span>
+				<form
+					method="post"
+					action={ retryNowURL(r.VideoID) }
+					hx-post={ string(retryNowURL(r.VideoID)) }
+					hx-target={ "#video-" + r.VideoID }
+					hx-swap="outerHTML"
+					class="card-nudge-form"
+				>
+					<button type="submit" class="btn-quiet" title="Summarize this video">Summarize</button>
+				</form>
+			} else {
+				// State 5: pending — discovered, not yet attempted; nudge button → summarize handler.
+				<span class="card-state muted">Not summarized</span>
+				<form
+					method="post"
+					action={ summarizeURL(r.VideoID) }
+					hx-post={ string(summarizeURL(r.VideoID)) }
+					hx-target={ "#video-" + r.VideoID }
+					hx-swap="outerHTML"
+					class="card-nudge-form"
+				>
+					<button type="submit" class="btn-quiet" title="Summarize this video">Summarize</button>
+				</form>
+			}
 		</div>
 	</li>
 }
-		</ul>
+
+// TapirSpinner is the summarizing animation: a Charmbracelet-style TUI panel —
+// three richly coloured ASCII tapir frames (inline span colours, snout wiggling
+// ∩→∪→~) cross-faded by CSS, plus a lipgloss-style progress bar whose mint fill
+// grows over the dim track. The panel is aria-hidden (decorative); the
+// "Summarizing…" label below carries the meaning for assistive tech.
+templ TapirSpinner() {
+	<div class="tapir-charm" aria-hidden="true">
+		<pre class="tapir-f1">@templ.Raw(tapirFrameHTML1)</pre>
+		<pre class="tapir-f2">@templ.Raw(tapirFrameHTML2)</pre>
+		<pre class="tapir-f3">@templ.Raw(tapirFrameHTML3)</pre>
+		<div class="tapir-bar"><span class="tapir-bar-fill" style={ "color:" + CharmMint }>{ tapirBarFill }</span></div>
+	</div>
+	<p class="tapir-label" role="status" aria-live="polite"><em>Summarizing…</em></p>
 }
+
+// processingCard is the in-flight summarization card. It replaces the Summarize
+// button card and polls /v/{id}/status every 2s, swapping itself (outerHTML, same
+// id as VideoCard) for whatever state comes back: it keeps polling while still
+// processing, and the summary/queued card it is eventually replaced by carries no
+// poll, so polling stops on its own when the fragment changes.
+templ processingCard(r store.SummaryRow) {
+	<li
+		class="card card-processing"
+		id={ "video-" + r.VideoID }
+		hx-get={ string(statusURL(r.VideoID)) }
+		hx-trigger="every 2s"
+		hx-swap="outerHTML"
+	>
+		<div class="card-title">{ displayTitle(r) }</div>
+		if cardMeta(r) != "" {
+			<div class="card-meta">{ cardMeta(r) }</div>
+		}
+		@TapirSpinner()
+	</li>
 }
 
 // DetailPage is the full summary view: text, highlights, takeaways, metadata,
@@ -116,6 +321,7 @@ templ summaryList(rows []store.SummaryRow) {
 templ DetailPage(r store.SummaryRow) {
 	@Layout("Tapir — " + displayTitle(r)) {
 		<article class="detail">
+			<p class="back"><a href="/">← Summaries</a></p>
 			<h1>{ displayTitle(r) }</h1>
 			<p class="meta">
 				if detailMeta(r) != "" {
@@ -141,20 +347,10 @@ templ DetailPage(r store.SummaryRow) {
 				<p class="source"><a href={ externalURL(r.URL) } rel="noopener noreferrer">watch on source ↗</a></p>
 			}
 			@ActionButtons(r.VideoID, actionSet(r.Actions))
-			<section>
+			// Lead with the attention-saving payload: Takeaways ("is this worth my
-				<h2>Summary</h2>
+			// time?") first, then Highlights, then the full Summary last (UX review
-				<p class="body">{ r.Summary }</p>
+			// A8). Takeaways/Highlights are conditional, so a video without them falls
-			</section>
+			// through to the Summary leading naturally.
-			if len(r.Highlights) > 0 {
-				<section>
-					<h2>Highlights</h2>
-					<ul>
-						for _, h := range r.Highlights {
-							<li>{ h }</li>
-						}
-					</ul>
-				</section>
-			}
 			if len(r.Takeaways) > 0 {
 				<section>
 					<h2>Takeaways</h2>
@@ -165,13 +361,27 @@ templ DetailPage(r store.SummaryRow) {
 					</ul>
 				</section>
 			}
+			if len(r.Highlights) > 0 {
+				<section>
+					<h2>Highlights</h2>
+					<ul>
+						for _, h := range r.Highlights {
+							<li>{ h }</li>
+						}
+					</ul>
+				</section>
+			}
+			<section>
+				<h2>Summary</h2>
+				<p class="body">{ r.Summary }</p>
+			</section>
 		</article>
 	}
 }
 
 // RegisterPage is the explicit registration step (ADR-012): an authenticated Dex
-// subject with no tapir user picks a display name and accepts the terms to create
+// subject with no tapir user picks a display name to create their account.
-// their account. errMsg, when set, reports a validation problem on the prior POST.
+// errMsg, when set, reports a validation problem on the prior POST.
 templ RegisterPage(email, errMsg string) {
 	@Layout("Tapir — Register") {
 		<article class="register">
@@ -188,10 +398,6 @@ templ RegisterPage(email, errMsg string) {
 					Display name
 					<input type="text" name="display_name" required autofocus/>
 				</label>
-				<label class="checkbox">
-					<input type="checkbox" name="accept_terms" value="yes" required/>
-					I accept the terms of use
-				</label>
 				<button type="submit" class="btn">Register</button>
 			</form>
 		</article>
@@ -202,7 +408,7 @@ templ RegisterPage(email, errMsg string) {
 // signed-in email, the user's connected video accounts (each with a Disconnect
 // control), a Connect-YouTube link when none is connected, and the delete-account
 // danger zone. flash surfaces a one-shot notification (disconnect/connect).
-templ AccountPage(displayName, email string, conns []store.Connection, flash string) {
+templ AccountPage(displayName, email string, conns []store.Connection, autoSummarize bool, channelErrors []store.ChannelError, flash string) {
 	@Layout("Tapir — Account") {
 		@flashBanner(flash)
 		<article class="account">
@@ -215,6 +421,34 @@ templ AccountPage(displayName, email string, conns []store.Connection, flash str
 					<dd>{ email }</dd>
 				}
 			</dl>
+			<section>
+				<h2>Summarization</h2>
+				<p class="muted">
+					Automatic summarizes new videos from about the last week as they are
+					discovered. Older videos stay browsable — summarize them on demand.
+					Manual lets you pick which videos to summarize — every new video appears
+					in your list with a Summarize button.
+				</p>
+				@summarizeModeControl(autoSummarize)
+			</section>
+			if len(channelErrors) > 0 {
+				<section class="channel-errors">
+					<h2>Unavailable channels</h2>
+					<p class="muted">
+						{ fmt.Sprintf("%d channel(s) returned errors on the last discovery pass.", len(channelErrors)) }
+						These may have been deleted or made private on YouTube.
+					</p>
+					<ul class="channel-error-list">
+						for _, ce := range channelErrors {
+							<li>
+								<span class="channel-error-name">{ ce.ChannelName }</span>
+								<span class="chip chip-warn">unavailable</span>
+								<span class="muted channel-error-since">since { ce.FirstSeen.Format("2006-01-02") }</span>
+							</li>
+						}
+					</ul>
+				</section>
+			}
 			<section>
 				<h2>Connected accounts</h2>
 				if len(conns) == 0 {
@@ -262,6 +496,26 @@ templ AccountPage(displayName, email string, conns []store.Connection, flash str
 	}
 }
 
+// summarizeModeControl is the auto/manual toggle, also returned standalone by
+// POST /account/summarize-mode (HTMX swaps it via outerHTML). The hidden field
+// submits the desired NEW value, so a single submit flips the mode; without JS the
+// form posts and the handler redirects back to /account.
+templ summarizeModeControl(auto bool) {
+	<div id="summarize-mode" class="summarize-mode">
+		<p>Current mode: <strong>{ summarizeModeLabel(auto) }</strong></p>
+		<form
+			method="post"
+			action="/account/summarize-mode"
+			hx-post="/account/summarize-mode"
+			hx-target="#summarize-mode"
+			hx-swap="outerHTML"
+		>
+			<input type="hidden" name="enabled" value={ boolStr(!auto) }/>
+			<button type="submit" class="btn-secondary">{ summarizeModeToggleLabel(auto) }</button>
+		</form>
+	</div>
+}
+
 // ActionButtons is the toggle group fragment returned by POST /v/{id}/action.
 // Each button submits its verb; HTMX swaps this element in place (outerHTML),
 // and without JS the form POSTs and the handler redirects back to the detail
@@ -276,20 +530,31 @@ templ ActionButtons(videoID string, active map[string]bool) {
 		hx-target="#action-buttons"
 		hx-swap="outerHTML"
 	>
-		for _, v := range actionVerbs {
+		// watched ↔ skipped are mutually exclusive (the store clears one when the
+		// other is set), so they read as a single segmented choice. "saved" is an
+		// independent toggle and sits apart (UX review C5).
+		<span class="segmented" role="group" aria-label="Watched or skipped">
+			@actionButton("watched", active["watched"])
+			@actionButton("skipped", active["skipped"])
+		</span>
+		@actionButton("saved", active["saved"])
+	</form>
+}
+
+// actionButton is one toggle button in the action group: a submit carrying its
+// verb, marked active (accent fill + ✓ prefix + aria-pressed) when currently set.
+templ actionButton(verb string, isActive bool) {
 	<button
 		type="submit"
 		name="action"
-				value={ v }
+		value={ verb }
-				class={ "action", templ.KV("active", active[v]) }
+		class={ "action", templ.KV("active", isActive) }
-				aria-pressed={ ariaPressed(active[v]) }
+		aria-pressed={ ariaPressed(isActive) }
 	>
-				if active[v] {
+		if isActive {
-					{ "✓ " + actionLabel(v) }
+			{ "✓ " + actionLabel(verb) }
 		} else {
-					{ actionLabel(v) }
+			{ actionLabel(verb) }
 		}
 	</button>
 }
-	</form>
-}

internal/web/welcome_test.go

@@ -0,0 +1,97 @@
+package web_test
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/web"
+)
+
+// fakeAuth is a configurable web.Auth for the landing-page tests: it reports a
+// fixed (user, ok) from CurrentUser and, when logged out, replicates DexAuth's
+// redirect split in Middleware — bare root → /welcome, deeper paths → login.
+// StubAuth can't express the logged-out case (it allows everything), so the
+// welcome routing needs this.
+type fakeAuth struct {
+	user web.User
+	ok   bool
+}
+
+func (f fakeAuth) CurrentUser(*http.Request) (web.User, bool) { return f.user, f.ok }
+func (f fakeAuth) Routes() http.Handler                       { return http.NewServeMux() }
+
+func (f fakeAuth) Middleware(h http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if f.ok {
+			h.ServeHTTP(w, r)
+			return
+		}
+		if r.URL.Path == "/" {
+			http.Redirect(w, r, "/welcome", http.StatusFound)
+			return
+		}
+		http.Redirect(w, r, "/auth/login", http.StatusFound)
+	})
+}
+
+// appWithAuth builds an App with a given Auth but no store wiring — enough for
+// the /welcome page (which never touches the store) and the unauthenticated
+// redirect paths (which never reach a handler).
+func appWithAuth(auth web.Auth) *web.App {
+	return &web.App{Auth: auth}
+}
+
+func TestWelcomeLoggedOut(t *testing.T) {
+	app := appWithAuth(fakeAuth{ok: false})
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/welcome", nil))
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+	require.Contains(t, html, "Get Started", "logged-out CTA present")
+	require.Contains(t, html, `href="/auth/login"`, "CTA links into the Dex flow")
+	require.NotContains(t, html, "Go to my Tapir", "no logged-in controls")
+	// Invites are owned by Authentik now (ADR-019); Tapir no longer handles
+	// invite links. The old "invite link will set up your account automatically"
+	// promise is stale and misleading — it must be gone.
+	require.NotContains(t, html, "invite link", "stale Authentik-superseded invite copy must be removed")
+	require.Contains(t, html, "invite-only", "honest invite-only framing present")
+}
+
+func TestWelcomeLoggedIn(t *testing.T) {
+	app := appWithAuth(fakeAuth{user: web.User{Subject: "s", Email: "me@d-ma.be"}, ok: true})
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/welcome", nil))
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	html := body(t, rec)
+	require.Contains(t, html, "Go to my Tapir", "logged-in CTA present")
+	require.Contains(t, html, `href="/"`, "links back into the app")
+	require.Contains(t, html, "me@d-ma.be", "greets by email")
+	require.NotContains(t, html, "Get Started", "no logged-out CTA")
+}
+
+func TestUnauthenticatedRootRedirectsToWelcome(t *testing.T) {
+	app := appWithAuth(fakeAuth{ok: false})
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+
+	require.Equal(t, http.StatusFound, rec.Code)
+	require.Equal(t, "/welcome", rec.Header().Get("Location"))
+}
+
+func TestUnauthenticatedDeepLinkRedirectsToLogin(t *testing.T) {
+	app := appWithAuth(fakeAuth{ok: false})
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/v/some-id", nil))
+
+	require.Equal(t, http.StatusFound, rec.Code)
+	require.Equal(t, "/auth/login", rec.Header().Get("Location"))
+}
+
+func TestAuthenticatedRootRendersList(t *testing.T) {
+	app := newApp(t)
+	resetDB(t, rawPool(t))
+	rec := do(t, app, httptest.NewRequest(http.MethodGet, "/", nil))
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Contains(t, body(t, rec), "<html", "authenticated root still renders the list page")
+}

test/acceptance/scenario_coverage_test.go

@@ -0,0 +1,212 @@
+package acceptance
+
+// This is the name-coverage gate for the BDD spec (see docs/use-cases/*.feature).
+// There is no godog runner — the .feature files are design records, and the real
+// behaviour is covered by the hand-written Go tests across the module. This test
+// keeps the two from drifting in the cheapest honest way: every non-@pending
+// Scenario must have an entry in scenarioCoverage pointing at a Go test that
+// actually exists. It does NOT prove the test exercises the scenario (only godog
+// could); it catches the common drift — "added a scenario, forgot the test", a
+// renamed/deleted covering test, or a scenario removed without cleaning the map.
+//
+// When you add a Scenario: either map it here to its covering test, or tag it
+// @pending in the .feature with a one-line reason for why it has no test yet.
+
+import (
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+	"testing"
+)
+
+// scenarioCoverage maps each non-@pending Scenario name to the Go test that
+// covers it. Keep it in sync with docs/use-cases/*.feature — the test below
+// fails if a scenario is unmapped, a mapped test is missing, or an entry no
+// longer matches a real non-pending scenario.
+var scenarioCoverage = map[string]string{
+	// ai_routing.feature
+	"Local AI produces the summary":                             "TestSummarize_LocalSucceeds",
+	"Local AI fails and the user has a BYO provider configured": "TestSummarize_FallsBackToBYO",
+	"Local AI fails and the user has no BYO provider":           "TestSummarize_LocalFailsNoBYO_NoExternalSend",
+	"A user without BYO never has content sent externally":      "TestSummarize_NoBYO_ContentOnlyLocal",
+
+	// landing_page.feature
+	"An unauthenticated visit to the root is sent to the welcome page":    "TestUnauthenticatedRootRedirectsToWelcome",
+	"The welcome page invites an unauthenticated visitor to start":        "TestWelcomeLoggedOut",
+	"An authenticated user on the welcome page sees their way in and out": "TestWelcomeLoggedIn",
+
+	// connect_account.feature
+	"Connect a YouTube account":                              "TestCallbackExchangesAndRecordsConnection",
+	"Connecting an account discovers videos immediately":     "TestCallbackTriggersDiscovery",
+	"Tokens are never stored in the clear":                   "TestCallbackExchangesAndRecordsConnection",
+	"Revoking a connection stops watching but keeps history": "TestDisconnectRemovesTokenAndConnectionKeepsAccount",
+
+	// summarize_mode.feature
+	"Auto mode summarizes recent new videos automatically":           "TestRunOnce_AutoMode_SkipsOldVideos",
+	"Auto mode lists older videos without summarizing them":          "TestRunOnce_AutoMode_SkipsOldVideos",
+	"Automatic is the default for a new user":                        "TestRegisteredUserDefaultsAutoSummarizeOn",
+	"Manual mode leaves new videos unsummarized":                     "TestRunOnce_ManualMode_SkipsUnrequested",
+	"Requesting a summary in manual mode queues it for the next run": "TestRunOnce_ManualMode_ProcessesRequested",
+
+	// registration.feature
+	"A new Dex subject is routed to registration":                               "TestUnregisteredSubjectRedirectedToRegister",
+	"Registering creates the account and its identity mapping":                  "TestRegisterCreatesExactlyOneUserAndIdentity",
+	"A returning subject passes straight through":                               "TestRegisteredSubjectPassesThrough",
+	"Deleting an account removes only my data and leaves other users untouched": "TestDeleteAccountWipesDataAndSecretsAndLogsOut",
+
+	// summarize_new_video.feature
+	"A subscribed channel posts a video that has captions":         "TestSubscribedVideoWithCaptionsIsSummarizedAndDelivered",
+	"A subscribed channel posts a video with no usable transcript": "TestVideoWithNoTranscriptIsSkipped",
+	"A channel I am not subscribed to posts a video":               "TestUnsubscribedChannelVideoIsNotProcessed",
+	"The same video is not summarized twice":                       "TestAlreadySummarizedVideoIsNotReprocessed",
+}
+
+var (
+	scenarioRe = regexp.MustCompile(`^\s*Scenario(?: Outline)?:\s*(.+?)\s*$`)
+	testFuncRe = regexp.MustCompile(`^func (Test\w+)\(`)
+)
+
+// scenario is one parsed Gherkin scenario and whether it is @pending.
+type scenario struct {
+	name    string
+	pending bool
+}
+
+func TestScenarioCoverage(t *testing.T) {
+	root := moduleRoot(t)
+
+	scenarios := parseScenarios(t, filepath.Join(root, "docs", "use-cases"))
+	if len(scenarios) == 0 {
+		t.Fatal("no scenarios parsed from docs/use-cases — wrong path?")
+	}
+	tests := allTestFuncNames(t, root)
+
+	// Index scenario names for the reverse (stale-entry) check.
+	active := map[string]bool{} // non-pending scenario names
+	var pending []string
+	for _, s := range scenarios {
+		if s.pending {
+			pending = append(pending, s.name)
+			continue
+		}
+		active[s.name] = true
+
+		// 1. Every non-pending scenario must be mapped.
+		fn, ok := scenarioCoverage[s.name]
+		if !ok {
+			t.Errorf("scenario %q has no coverage entry — map it in scenarioCoverage to a covering test, or tag it @pending in the .feature", s.name)
+			continue
+		}
+		// 2. The mapped test must actually exist.
+		if !tests[fn] {
+			t.Errorf("scenario %q maps to %q, which is not a Test function anywhere in the module", s.name, fn)
+		}
+	}
+
+	// 3. No stale entries: every map key must be a real, non-pending scenario.
+	for name := range scenarioCoverage {
+		if !active[name] {
+			t.Errorf("scenarioCoverage has entry %q, which is not a current non-pending scenario (renamed, removed, or now @pending?)", name)
+		}
+	}
+
+	if len(pending) > 0 {
+		t.Logf("%d @pending scenario(s) without a test (tracked, not required): %s",
+			len(pending), strings.Join(pending, "; "))
+	}
+}
+
+// parseScenarios reads every *.feature under dir and returns its scenarios with
+// their @pending status. A scenario is @pending when a `@pending` tag line
+// precedes it (tags survive intervening comment lines, the layout these files
+// use); the flag is consumed at the Scenario line and reset afterward.
+func parseScenarios(t *testing.T, dir string) []scenario {
+	t.Helper()
+	entries, err := os.ReadDir(dir)
+	if err != nil {
+		t.Fatalf("read use-cases dir: %v", err)
+	}
+	var out []scenario
+	for _, e := range entries {
+		if e.IsDir() || !strings.HasSuffix(e.Name(), ".feature") {
+			continue
+		}
+		b, err := os.ReadFile(filepath.Join(dir, e.Name()))
+		if err != nil {
+			t.Fatalf("read %s: %v", e.Name(), err)
+		}
+		pending := false
+		for _, line := range strings.Split(string(b), "\n") {
+			trimmed := strings.TrimSpace(line)
+			if strings.HasPrefix(trimmed, "@") {
+				if strings.Contains(trimmed, "@pending") {
+					pending = true
+				}
+				continue
+			}
+			if m := scenarioRe.FindStringSubmatch(line); m != nil {
+				out = append(out, scenario{name: m[1], pending: pending})
+				pending = false
+			}
+			// comment (#) and step lines leave a set @pending intact until the
+			// scenario consumes it; a blank line between scenarios is harmless.
+		}
+	}
+	return out
+}
+
+// allTestFuncNames walks the module for `func TestXxx(` declarations, excluding
+// this file (whose regex literal would otherwise look like a definition).
+func allTestFuncNames(t *testing.T, root string) map[string]bool {
+	t.Helper()
+	self := "scenario_coverage_test.go"
+	names := map[string]bool{}
+	err := filepath.WalkDir(root, func(path string, d os.DirEntry, err error) error {
+		if err != nil {
+			return err
+		}
+		if d.IsDir() {
+			if d.Name() == ".git" {
+				return filepath.SkipDir
+			}
+			return nil
+		}
+		if !strings.HasSuffix(path, "_test.go") || filepath.Base(path) == self {
+			return nil
+		}
+		b, err := os.ReadFile(path)
+		if err != nil {
+			return err
+		}
+		for _, line := range strings.Split(string(b), "\n") {
+			if m := testFuncRe.FindStringSubmatch(line); m != nil {
+				names[m[1]] = true
+			}
+		}
+		return nil
+	})
+	if err != nil {
+		t.Fatalf("walk module: %v", err)
+	}
+	return names
+}
+
+// moduleRoot walks up from the working directory to the dir containing go.mod.
+func moduleRoot(t *testing.T) string {
+	t.Helper()
+	dir, err := os.Getwd()
+	if err != nil {
+		t.Fatalf("getwd: %v", err)
+	}
+	for {
+		if _, err := os.Stat(filepath.Join(dir, "go.mod")); err == nil {
+			return dir
+		}
+		parent := filepath.Dir(dir)
+		if parent == dir {
+			t.Fatal("go.mod not found walking up from cwd")
+		}
+		dir = parent
+	}
+}