feat(report): Stage-0 usage gate counts from a baseline date (default 2026-06-11)

The return-usage gate (ADR-016) counted distinct active weeks over ALL history,
so pre-launch noise — testing churn and the period the pilot sat blocked on zero
summaries — would inflate the signal. Add a baseline: ActiveWeeks(ctx, since)
filters login_events + summary_actions to seen_at/acted_at >= since. The report
command sets it to TAPIR_USAGE_GATE_START (YYYY-MM-DD, default 2026-06-11 — the
morning the pilot was unblocked) and prints the baseline. The gate now measures
whether users RETURN once it genuinely works.

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

DECISIONS.md

@@ -827,10 +827,18 @@ The fix is resilience around it, not replacing it.
 **Decision.**
 1. **Ordered endpoint chain (`summarizer.NewChain`).** Endpoints are tried in order; the first to
    return a *parseable* summary wins. Default chain:
-   `koala/phi4-mini` (primary, local) → `koala/phi4-14b` (fallback, local) →
+   `koala/phi4-mini` (primary, local) → `iguana/gemma4-26b` (fallback, local on a
-   `berget/mistral-small` (worst-case, external). All three are reached through the **one** LiteLLM
+   *different host*) → `berget/mistral-small` (worst-case, external). All three are reached through
-   gateway by alias — the gateway already fronts both llama-swap and berget — so a fallback is a
+   the **one** LiteLLM gateway by alias — the gateway already fronts both llama-swap and berget — so
-   different alias, not a second client config.
+   a fallback is a different alias, not a second client config.
+
+   **Update 2026-06-11:** the local fallback moved from `koala/phi4-14b` to `iguana/gemma4-26b`.
+   koala now carries other GPU loads, so keeping the fallback on koala competed with them; iguana
+   (M2 Ultra) has the headroom, and a different host is also a different egress IP for the rare
+   fallback fetch. `gemma4-26b` is the brain-validated homelab general-purpose model (agentsquad
+   H2/H3 executor) and returned valid summary JSON on the real prompt in a smoke test
+   (~37s incl. cold-load — fine for a path hit only when the fast primary fails). Pure config:
+   `TAPIR_FALLBACK_MODEL`.
 2. **A parse failure advances the chain, same as a transport error.** "Reliably summarized" means
    *parseable summary returned*, not *HTTP 200*. This is the behaviour the old Primary→Fallback
    shape missed.
@@ -958,6 +966,168 @@ schema change (reuses `transcript_status` from migration 007).
 
 ---
 
+## ADR-026 — Foreground caption fetches take priority; the credentials probe is dead
+
+**Status:** Accepted (2026-06-10). **Pillar A of the manual-mode UX work** (Pillar B was
+ADR-025). Builds on ADR-014 (the shared per-IP gate).
+
+**Context.** Every caption fetch — the background sweep and the web click-path — shared one
+process-wide rate gate equally. So a user waiting on a "Summarize" click competed with the
+firehose for both pacing and the scarce pre-429 window; on a busy IP the click was slow or
+429'd while the background churned.
+
+**Decision.** A context-marked priority lane. The web path
+(`engineProcessor.ProcessVideo`) wraps its context with `ForegroundContext`; the gate gives
+foreground fetches a token immediately, while **background fetches yield** — they wait until no
+foreground fetch is pending before taking a token. Threaded via a context value (not new
+signatures) and a process-wide `foregroundPending` counter. Clicks are rare and bursty, so the
+background barely loses throughput; the waiting human gets the next (and cleanest) slot.
+
+**Credentials probe — rejected, not built.** The idea was to fetch captions with the user's
+auth in manual mode to dodge 429s. It is a dead end, already settled by ADR-010 and the code:
+the caption path is *deliberately anonymous* because the InnerTube/timedtext endpoints **reject
+or break on authenticated requests** (`captions.go`: "no OAuth token — it can break the
+timedtext endpoint"). The user's OAuth (a Data API credential) does not authenticate InnerTube
+at all, and the official `captions.download` is owner-only (403 on third-party). So auth cannot
+help here and can actively hurt. No probe needed — building one would only re-confirm the ADR.
+
+**Reversibility.** Context-marker + a yield loop in the gate; removing the marker collapses to
+the prior equal-share behaviour. No schema or API change.
+
+---
+
+## ADR-027 — Chat with a video's stored transcript (deeper-dive, on an already-summarized video)
+
+**Status:** Accepted (2026-06-11). **Consumes ADR-021** (the shared, video-keyed transcript
+store) for the first time beyond summarization; **uses the ADR-022 chain models**; relates to
+ADR-012 (isolation) and ADR-016 (the Stage-0 gate).
+
+**Context — observed demand, not hypothetical.** The maintainer read 10+ real pilot summaries
+and reported the reactions: *many good; some he wanted to dig deeper into; some less useful*
+(the "less useful" split between weak-*model* output and uninteresting-*video* content). The
+middle reaction is the signal: a good summary that makes the reader want *more* is the summary
+succeeding at triage and then hitting a wall — there is nowhere to go deeper short of watching
+the video. That want is the feature. It is also the cheapest possible feature to satisfy
+honestly, because ADR-021 already persists the transcript: the deeper-dive runs entirely on
+stored public-content text + local models, touching **no** caption fetch and **no** YouTube.
+
+**Decision.** Add a per-video chat that lets the user ask questions against a video's stored
+transcript.
+1. **Entry from the summary view only.** A "dig deeper / ask" affordance on a summarized video —
+   the chat lives exactly where the "I want more" reaction happens. No standalone chat surface.
+2. **Stored-transcript-only (load-bearing constraint).** Chat is available **only** for videos
+   that already have a stored transcript. It never triggers a caption fetch, so it cannot touch
+   the rate gate, the 429 surface, or YouTube at all — the entire account-safety constraint that
+   governs the rest of Tapir is satisfied *by construction* here, not by careful gating. (Entry
+   being "from a summary" guarantees the transcript exists.) On-demand fetch for un-stored videos
+   is explicitly deferred.
+3. **Model = the summary's model by default; user-switchable among the ADR-022 chain models**
+   (`phi4-mini` / `gemma4-26b` / `mistral-small` to start). This is deliberate: it doubles as
+   live model-comparison instrumentation — ask the same question of the same transcript under two
+   models and the difference is directly felt. This is the mechanism by which the maintainer
+   learns *which* model is worth defaulting to, and it is the multi-model-analysis direction
+   ADR-021 anticipated, arriving as a user-facing capability.
+   - Chat is a **read-bounded retrieval/QA task** (the user supplies the focus), which is
+     *easier* than summarization (the model must decide what matters). So a model that summarizes
+     mediocrely may chat well — chat is plausibly a partial remedy for the weak-summary case, not
+     an inheritor of it.
+4. **Ephemeral chat (v1).** No persisted history; chat is per-session. Persisting per-user,
+   RLS-scoped history is deferred until there is evidence anyone wants to revisit a conversation.
+5. **Chat-only, trust-the-model (v1) — with a recorded limitation.** The chat does not expose the
+   raw transcript for verification in v1 (kept simple). **Known limitation:** because some
+   summaries were weak-model output, the user has reason not to fully trust a chat answer's
+   fidelity to the transcript, and v1 gives no in-UI way to check. The model-switcher partially
+   compensates (two models disagreeing on the same question is itself a signal). A
+   "show source / view transcript" verification path is the natural **v2** and is *not*
+   foreclosed — ADR-021's stored transcript already makes it cheap. Recorded so v2 is a known
+   next step, not a rediscovery.
+
+**Why this is safe and in-scope.** It adds no caption-fetch surface (stored-only), no new
+non-RLS table (transcripts already shared per ADR-021; ephemeral chat stores nothing), and no
+auth change. It is additive to the read path. The one genuine product expansion — Tapir becomes
+an interactive transcript-QA tool, not only a summarizer — is justified by *observed* demand from
+real reading, which is exactly the kind of evidence the Stage-0 discipline asks for before
+building.
+
+**Relation to the Stage-0 gate.** This is **not** a return-nudge (those stay deferred, ADR-020) —
+it adds nothing that prompts the user to return; it deepens the value *once they are already
+reading*. It does not contaminate the unprompted-return signal. If anything it strengthens the
+"useful to me" case the gate measures, by giving a good summary somewhere to lead.
+
+**Reversibility.** Additive read-path feature over the unchanged engine + the ADR-021 store.
+Removing the summary-view affordance removes the feature; nothing else depends on it. Ephemeral =
+no migration, no stored state to unwind. Spec: `docs/specs/chat-with-transcript.md`.
+
+---
+
+## ADR-028 — Onboarding burst: pick likely-good videos, summarize them with a stronger model
+
+**Status:** Accepted (2026-06-11). **Refines ADR-018** (the connect-time burst) and **ADR-020**
+(recency-bounded auto-summarize). **Builds on ADR-022** (the endpoint chain), **ADR-023**
+(the discovery-time `videos.list` enrichment), and **ADR-021** (the shared transcript cache).
+Triggered by a Phase-1 investigation of the live pilot DB.
+
+**Context.** A new user's first session decides whether they return (the Stage-0 gate, ADR-016).
+The connect-time burst (ADR-018: summarize ≤`TAPIR_ONBOARD_SUMMARIZE_COUNT` newest videos so the
+feed isn't empty) *fires* in production, but a live-DB investigation of the second pilot user
+("Jonte") found it delivers a weak first impression for two reasons, and ruled out a third idea:
+
+1. **Junk picks.** Selection was pure newest-first (`NewestUnsummarizedVideoIDs`,
+   `ORDER BY published_at DESC`) with **no quality signal**. Jonte's live burst-3 were a
+   stock-ticker **livestream** + two regional news clips — newest, not best. The cheap signals
+   that *could* gate this (duration, live status) are fetched by ADR-023's `videos.list`
+   enrichment at discovery and then **thrown away**: the `videos.duration_s` column (migration
+   001) was never written.
+2. **Weakest model on the first impression.** All burst summaries ran on `koala/phi4-mini` — the
+   documented weak link (ADR-022 was born from its failures). The stronger, brain-validated
+   `iguana/gemma4-26b` was never used, even though the burst is only ~3 summaries.
+3. **Cached-first instant summaries — REJECTED.** The idea: skip the fetch, summarize
+   already-cached transcripts (ADR-021) instantly. The pilot numbers kill it — only **11 videos**
+   overlap between the two users (~3% of each ~350–400-video library), **0** cached-and-
+   unsummarized, and a new user's newest-20 unsummarized are **20/20 NOT cached**. Newest-first
+   and cached-first are structurally incompatible: fresh uploads are exactly what nobody has
+   fetched. An empty lever at pilot scale.
+
+**Decision.**
+1. **Persist `duration_s` at discovery.** `filterLowValue` (ADR-023) already has each candidate's
+   duration in hand; carry it onto the kept `domain.Video` and have `UpsertVideo` write it,
+   COALESCE-preserving a known value (the channel-title backfill stance, migration 014). No new
+   migration — the column exists. The connect-triggered discovery pass runs *before* the burst,
+   so a fresh user's candidates are enriched in time.
+2. **Junk-avoiding selection.** A new `OnboardBurstVideoIDs(userID, limit, minSeconds, maxSeconds)`
+   keeps the newest-first order but drops a video when its duration is *known* and outside
+   `[minSeconds, maxSeconds]` — `minSeconds` = `TAPIR_MIN_VIDEO_SECONDS` (60, the Shorts floor),
+   `maxSeconds` = new `TAPIR_ONBOARD_MAX_VIDEO_SECONDS` (default 14400 = 4h, to drop multi-hour
+   livestream VODs that pass the live filter once ended). A NULL duration is **unknown** — kept
+   (degrade-open) but ranked after known-good rows. **has-captions stays un-gateable pre-fetch**
+   (only knowable after a gate fetch or a ~0-probability cache hit); selection only *avoids
+   known-junk*, it does not *promise* captions.
+3. **Stronger model for the burst only.** `TAPIR_ONBOARD_SUMMARIZER_MODEL` (default
+   `iguana/gemma4-26b`) leads a burst-specific summarizer chain (onboard model first, then the
+   standard ADR-022 chain as resilience, deduped), wrapped in a burst-specific processor over the
+   *same* store/cache/sink — a pure wiring choice; the engine and ports are unchanged (ADR-003).
+   Empty or equal-to-primary collapses the burst back onto the shared processor.
+
+**Not a throughput change.** The caption rate gate (ADR-014) and the foreground priority lane
+(ADR-026) are untouched — same pacing, same cap. This changes *which* ≤3 videos the burst spends
+its fetches on and *which model* summarizes them, never how fast or how many. The engine's
+existing read-stored-first (ADR-021) is unchanged and still yields a free instant summary on the
+rare cache hit — we simply do not *select* for cache hits.
+
+**Consequences.** Better odds of a strong first session: the burst avoids the obvious junk and
+runs the better model on the one impression that decides return. The selection improvement is
+forward-looking — existing rows have NULL `duration_s` until their next discovery pass backfills
+it (lazy, like channel_title); a brand-new user benefits immediately because connect-discovery
+runs first. `duration_s` becoming live also unblocks future length-aware features (feed sorting,
+"long read" badges) for free.
+
+**Reversibility.** Pure config + wiring + one column write + one query, no migration.
+`TAPIR_ONBOARD_MAX_VIDEO_SECONDS=0` (and `TAPIR_MIN_VIDEO_SECONDS=0`) restores pure newest-first;
+`TAPIR_ONBOARD_SUMMARIZER_MODEL=""` collapses the burst back to the shared processor.
+Spec: `docs/specs/onboarding-wow-burst.md`.
+
+---
+
 ## Rejected alternatives
 
 Approaches considered during the 2026-06-02 planning + grill session and **deliberately not
@@ -981,6 +1151,7 @@ maps to the ADR that settles it.
 | 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 |
+| Cached-transcript-first onboarding burst (instant, zero-fetch picks) | Live pilot DB: ~3% cross-user video overlap, 0 cached-and-unsummarized, a new user's newest-20 are 20/20 uncached — newest-first and cached-first are structurally incompatible. Empty lever at pilot scale | ADR-028 |
 
 If a future case genuinely reopens one of these, that's a new ADR superseding the relevant one —
 not a silent reversal.

cmd/tapir/chat_test.go

@@ -0,0 +1,64 @@
+package main
+
+import (
+	"strings"
+	"testing"
+
+	"gitea.d-ma.be/mathias/tapir/internal/config"
+)
+
+// TestChatModelsReuseTheChainLocalFirst: the switcher offers the ADR-022 chain in
+// order, deduped — primary, local fallback, cloud.
+func TestChatModelsReuseTheChainLocalFirst(t *testing.T) {
+	got := chatModels(config.Config{
+		SummarizerModel:    "koala/phi4-mini",
+		FallbackModel:      "iguana/gemma4-26b",
+		CloudFallbackModel: "berget/mistral-small",
+	})
+	want := []string{"koala/phi4-mini", "iguana/gemma4-26b", "berget/mistral-small"}
+	if strings.Join(got, ",") != strings.Join(want, ",") {
+		t.Fatalf("chatModels = %v, want %v", got, want)
+	}
+}
+
+// TestChatCloudModelAbsentWhenDisabled: the local-first / NDA lever — with the
+// cloud fallback empty (TAPIR_CLOUD_FALLBACK_MODEL=""), no external model is
+// offered in the switcher, so chat content never leaves the local stack (ADR-027,
+// honouring ADR-022's "content stays local" guarantee).
+func TestChatCloudModelAbsentWhenDisabled(t *testing.T) {
+	got := chatModels(config.Config{
+		SummarizerModel:    "koala/phi4-mini",
+		FallbackModel:      "iguana/gemma4-26b",
+		CloudFallbackModel: "",
+	})
+	for _, m := range got {
+		if strings.HasPrefix(m, "berget/") || strings.Contains(m, "mistral") {
+			t.Fatalf("cloud model %q offered though the cloud fallback is disabled", m)
+		}
+	}
+	want := []string{"koala/phi4-mini", "iguana/gemma4-26b"}
+	if strings.Join(got, ",") != strings.Join(want, ",") {
+		t.Fatalf("chatModels = %v, want %v", got, want)
+	}
+}
+
+// TestChatModelsDedup: a config that reuses one alias across slots collapses to a
+// single switcher entry (no duplicate options).
+func TestChatModelsDedup(t *testing.T) {
+	got := chatModels(config.Config{
+		SummarizerModel:    "koala/phi4-mini",
+		FallbackModel:      "koala/phi4-mini",
+		CloudFallbackModel: "",
+	})
+	if len(got) != 1 || got[0] != "koala/phi4-mini" {
+		t.Fatalf("chatModels = %v, want a single deduped entry", got)
+	}
+}
+
+// TestBuildChatNilWithoutGateway: no gateway → no chat backend (routes unmounted,
+// read path unaffected).
+func TestBuildChatNilWithoutGateway(t *testing.T) {
+	if c := buildChat(config.Config{GatewayURL: ""}); c != nil {
+		t.Fatal("buildChat must return nil without a gateway URL")
+	}
+}

cmd/tapir/main.go

@@ -197,6 +197,15 @@ func cmdServe(ctx context.Context, log *slog.Logger) error {
 	secretStore := secrets.NewFileStore(cfg.SecretsFile)
 	app := &web.App{Store: st, Identity: st, Auth: authn, Secrets: secretStore, Log: log, RecencyWindow: cfg.AutoSummarizeWindow}
 
+	// Per-video deeper-dive chat over the STORED transcript (ADR-027). Enabled
+	// whenever a gateway is configured — it needs no YouTube credentials because it
+	// never fetches. Guarded so a typed-nil never lands in the interface field
+	// (which would mount the routes over a nil backend).
+	if c := buildChat(cfg); c != nil {
+		app.Chat = c
+		log.Info("web chat enabled (stored-transcript only)", "models", chatModels(cfg))
+	}
+
 	// 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.
@@ -253,23 +262,36 @@ func cmdServe(ctx context.Context, log *slog.Logger) error {
 		// 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)
-		// Onboarding burst (Feature 1): after the connect-triggered discovery pass,
+		// Onboarding burst (Feature 1, refined by ADR-028): after the connect-triggered
-		// summarize up to OnboardSummarizeCount of the user's NEWEST unsummarized
+		// discovery pass, summarize up to OnboardSummarizeCount of the user's newest
-		// videos so a fresh account gets real summaries in its first session. Hard
+		// LIKELY-GOOD unsummarized videos so a fresh account gets a strong first
-		// cap; explicit, so it bypasses the recency window — but every fetch still
+		// session. Selection avoids known-junk (Shorts/over-long/livestream VODs via
-		// goes through globalFetchGate via the Processor. No-op when disabled
+		// the persisted duration); the burst leads its chain with the stronger onboard
-		// (count 0) or queue-only (no Processor).
+		// model. Hard cap; explicit, so it bypasses the recency window — but every
+		// fetch still goes through globalFetchGate. No-op when disabled (count 0) or
+		// queue-only (no processor).
+		//
+		// burstProcessor leads with the stronger model (ADR-028); it collapses onto the
+		// shared Processor when the onboard model is empty/equal-to-primary or the
+		// engine config is incomplete.
+		burstProcessor := app.Processor
+		if burstEngine, berr := buildBurstProcessor(cfg, st); berr != nil {
+			return berr
+		} else if burstEngine != nil {
+			burstProcessor = &engineProcessor{engine: burstEngine, store: st}
+			log.Info("onboarding burst uses a stronger model", "onboard_model", cfg.OnboardSummarizerModel)
+		}
 		onboard := func(ctx context.Context, userID string) {
-			if cfg.OnboardSummarizeCount <= 0 || app.Processor == nil {
+			if cfg.OnboardSummarizeCount <= 0 || burstProcessor == nil {
 				return
 			}
-			ids, err := st.NewestUnsummarizedVideoIDs(ctx, userID, cfg.OnboardSummarizeCount)
+			ids, err := st.OnboardBurstVideoIDs(ctx, userID, cfg.OnboardSummarizeCount, cfg.MinVideoSeconds, cfg.OnboardMaxVideoSeconds)
 			if err != nil {
-				log.Warn("onboarding: list newest unsummarized", "user", userID, "err", err)
+				log.Warn("onboarding: list burst candidates", "user", userID, "err", err)
 				return
 			}
 			for _, id := range ids {
-				if err := app.Processor.ProcessVideo(ctx, userID, id); err != nil {
+				if err := burstProcessor.ProcessVideo(ctx, userID, id); err != nil {
 					log.Warn("onboarding: summarize", "user", userID, "video", id, "err", err)
 				}
 			}

cmd/tapir/processor.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"strings"
 
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/chat"
 	"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"
@@ -51,13 +52,7 @@ func (f videoFetcher) FetchVideo(ctx context.Context, userID, videoID string) (d
 // alias, not a second client config. Empty model entries are skipped, so a
 // client deployment can set the cloud fallback empty to keep content local.
 func buildSummarizer(cfg config.Config) *summarizer.Summarizer {
-	mk := func(model string) summarizer.Endpoint {
+	mk := summarizerEndpoint(cfg)
-		return summarizer.Endpoint{
-			Client:   llm.New(cfg.GatewayURL, cfg.GatewayKey, model, cfg.SummarizerTimeout, llm.WithMaxTokens(cfg.SummaryMaxTokens)),
-			Provider: providerOf(model),
-			Model:    model,
-		}
-	}
 	eps := []summarizer.Endpoint{mk(cfg.SummarizerModel)}
 	if cfg.FallbackModel != "" && cfg.FallbackModel != cfg.SummarizerModel {
 		eps = append(eps, mk(cfg.FallbackModel))
@@ -68,6 +63,98 @@ func buildSummarizer(cfg config.Config) *summarizer.Summarizer {
 	return summarizer.NewChain(eps, cfg.MaxTranscriptChars)
 }
 
+// summarizerEndpoint returns a constructor for a chain endpoint over the one
+// LiteLLM gateway, varying only the model alias (the gateway fronts both
+// llama-swap and berget). Shared by the standard and burst chains.
+func summarizerEndpoint(cfg config.Config) func(model string) summarizer.Endpoint {
+	return func(model string) summarizer.Endpoint {
+		return summarizer.Endpoint{
+			Client:   llm.New(cfg.GatewayURL, cfg.GatewayKey, model, cfg.SummarizerTimeout, llm.WithMaxTokens(cfg.SummaryMaxTokens)),
+			Provider: providerOf(model),
+			Model:    model,
+		}
+	}
+}
+
+// burstChainModels is the ordered, deduped model list for the onboarding burst
+// (ADR-028): the stronger onboard model leads, then the standard ADR-022 chain
+// (primary → local fallback → cloud) follows as resilience. Empty entries are
+// dropped and duplicates collapsed, so the NDA lever (empty cloud fallback) keeps
+// the burst chain fully local exactly as the standard chain does.
+func burstChainModels(cfg config.Config) []string {
+	var models []string
+	add := func(m string) {
+		if m == "" {
+			return
+		}
+		for _, e := range models {
+			if e == m {
+				return
+			}
+		}
+		models = append(models, m)
+	}
+	add(cfg.OnboardSummarizerModel)
+	add(cfg.SummarizerModel)
+	add(cfg.FallbackModel)
+	add(cfg.CloudFallbackModel)
+	return models
+}
+
+// buildBurstSummarizer builds the onboarding-burst summarizer chain (ADR-028):
+// the onboard model first, then the standard chain as fallback, deduped.
+func buildBurstSummarizer(cfg config.Config) *summarizer.Summarizer {
+	mk := summarizerEndpoint(cfg)
+	var eps []summarizer.Endpoint
+	for _, m := range burstChainModels(cfg) {
+		eps = append(eps, mk(m))
+	}
+	return summarizer.NewChain(eps, cfg.MaxTranscriptChars)
+}
+
+// chatModels is the ordered, local-first set of models offered in the chat
+// switcher (ADR-027), reusing the ADR-022 chain: primary → local fallback →
+// cloud. Empty entries are dropped and duplicates collapsed, so a client/NDA
+// deployment that sets the cloud fallback empty simply has no external model in
+// the switcher — the same local-first lever the summarizer honours.
+func chatModels(cfg config.Config) []string {
+	var models []string
+	add := func(m string) {
+		if m == "" {
+			return
+		}
+		for _, e := range models {
+			if e == m {
+				return
+			}
+		}
+		models = append(models, m)
+	}
+	add(cfg.SummarizerModel)
+	add(cfg.FallbackModel)
+	add(cfg.CloudFallbackModel)
+	return models
+}
+
+// buildChat wires the per-video chat service (ADR-027): a Completer factory over
+// the SAME LiteLLM gateway the summarizer uses (a different alias per model, not a
+// second client config) and the same transcript-truncation budget. It returns nil
+// when no gateway is configured — chat is simply not mounted, the read path is
+// unaffected. It deliberately takes NO YouTube source: chat is stored-only.
+func buildChat(cfg config.Config) *chat.Service {
+	if cfg.GatewayURL == "" {
+		return nil
+	}
+	models := chatModels(cfg)
+	if len(models) == 0 {
+		return nil
+	}
+	newClient := func(model string) chat.Completer {
+		return llm.New(cfg.GatewayURL, cfg.GatewayKey, model, cfg.SummarizerTimeout, llm.WithMaxTokens(cfg.SummaryMaxTokens))
+	}
+	return chat.New(newClient, models, cfg.MaxTranscriptChars)
+}
+
 // providerOf maps a model alias to the domain AIProvider recorded on summaries.
 // A "berget/" alias is an external provider; everything else is the local stack.
 func providerOf(model string) string {
@@ -82,15 +169,7 @@ func buildProcessor(cfg config.Config, st *store.Store) (*usecase.Engine, error)
 		return nil, nil
 	}
 
-	secretStore := secrets.NewFileStore(cfg.SecretsFile)
+	src := newYouTubeSource(cfg, secrets.NewFileStore(cfg.SecretsFile))
-	src := youtube.New(youtube.Config{
-		ClientID:           cfg.YTClientID,
-		ClientSecret:       cfg.YTClientSecret,
-		TokenSecretRef:     cfg.YTTokenRef,
-		PreferredLanguages: []string{"en"},
-		MinVideoSeconds:    cfg.MinVideoSeconds,
-	}, secretStore)
-
 	sum := buildSummarizer(cfg)
 
 	// The store is both the summary sink and the shared transcript cache (ADR-021):
@@ -101,6 +180,38 @@ func buildProcessor(cfg config.Config, st *store.Store) (*usecase.Engine, error)
 	return eng, nil
 }
 
+// newYouTubeSource builds the captions-first VideoSource shared by the standard
+// and burst processors — same per-process YouTube credentials and ADR-023 Shorts
+// filter; only the summarizer chain differs between them.
+func newYouTubeSource(cfg config.Config, secretStore ports.SecretStore) ports.VideoSource {
+	return youtube.New(youtube.Config{
+		ClientID:           cfg.YTClientID,
+		ClientSecret:       cfg.YTClientSecret,
+		TokenSecretRef:     cfg.YTTokenRef,
+		PreferredLanguages: []string{"en"},
+		MinVideoSeconds:    cfg.MinVideoSeconds,
+	}, secretStore)
+}
+
+// buildBurstProcessor wires a processor whose summarizer leads with the stronger
+// onboard model (ADR-028), used only by the connect-time burst over the SAME
+// store / transcript cache / sink — a wiring choice; the engine and ports are
+// unchanged. Returns (nil, nil) — the collapse lever — when the onboard model is
+// empty or equal to the primary (the burst then reuses the shared processor), or
+// when the engine config is incomplete (queue-only, same as buildProcessor).
+func buildBurstProcessor(cfg config.Config, st *store.Store) (*usecase.Engine, error) {
+	if cfg.OnboardSummarizerModel == "" || cfg.OnboardSummarizerModel == cfg.SummarizerModel {
+		return nil, nil
+	}
+	if cfg.GatewayURL == "" || cfg.YTClientID == "" || cfg.YTClientSecret == "" || cfg.SecretsFile == "" {
+		return nil, nil
+	}
+	src := newYouTubeSource(cfg, secrets.NewFileStore(cfg.SecretsFile))
+	eng := usecase.NewEngine(src, buildBurstSummarizer(cfg), st)
+	eng.Transcripts = st
+	return eng, 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
@@ -113,6 +224,11 @@ type engineProcessor struct {
 }
 
 func (p *engineProcessor) ProcessVideo(ctx context.Context, userID, videoID string) error {
+	// This is the user-initiated (foreground) path — a click on "Summarize",
+	// "Try now", or a pasted URL. Mark the context so the caption gate gives it
+	// priority over the background sweep (ADR-026, Pillar A).
+	ctx = youtube.ForegroundContext(ctx)
+
 	row, err := p.store.GetVideoRow(ctx, userID, videoID)
 	if err != nil {
 		return fmt.Errorf("load video %q: %w", videoID, err)

cmd/tapir/processor_test.go

@@ -45,3 +45,65 @@ func TestBuildProcessorNilOnIncompleteConfig(t *testing.T) {
 		})
 	}
 }
+
+// TestBurstChainModelsLeadsWithOnboardModel: the onboarding burst chain (ADR-028)
+// leads with the stronger onboard model, then falls back through the standard
+// ADR-022 chain (primary -> local fallback -> cloud), deduped.
+func TestBurstChainModelsLeadsWithOnboardModel(t *testing.T) {
+	got := burstChainModels(config.Config{
+		OnboardSummarizerModel: "iguana/gemma4-26b",
+		SummarizerModel:        "koala/phi4-mini",
+		FallbackModel:          "iguana/gemma4-26b", // also the onboard model -> dedup
+		CloudFallbackModel:     "berget/mistral-small",
+	})
+	want := []string{"iguana/gemma4-26b", "koala/phi4-mini", "berget/mistral-small"}
+	if len(got) != len(want) {
+		t.Fatalf("burstChainModels = %v, want %v", got, want)
+	}
+	for i := range want {
+		if got[i] != want[i] {
+			t.Fatalf("burstChainModels = %v, want %v", got, want)
+		}
+	}
+}
+
+// TestBurstChainModelsCloudAbsentWhenDisabled: the NDA lever holds for the burst
+// too — empty cloud fallback keeps the burst chain fully local.
+func TestBurstChainModelsCloudAbsentWhenDisabled(t *testing.T) {
+	got := burstChainModels(config.Config{
+		OnboardSummarizerModel: "iguana/gemma4-26b",
+		SummarizerModel:        "koala/phi4-mini",
+		CloudFallbackModel:     "",
+	})
+	for _, m := range got {
+		if m == "" || m == "berget/mistral-small" {
+			t.Fatalf("cloud model leaked into burst chain: %v", got)
+		}
+	}
+}
+
+// TestBuildBurstProcessorNilWhenCollapsed: an empty or primary-equal onboard model
+// collapses the burst onto the shared processor (buildBurstProcessor returns nil).
+func TestBuildBurstProcessorNilWhenCollapsed(t *testing.T) {
+	base := config.Config{
+		GatewayURL:      "http://gw/v1",
+		YTClientID:      "id",
+		YTClientSecret:  "secret",
+		SecretsFile:     "/tmp/secrets.json",
+		SummarizerModel: "koala/phi4-mini",
+	}
+	t.Run("empty onboard model", func(t *testing.T) {
+		base.OnboardSummarizerModel = ""
+		eng, err := buildBurstProcessor(base, nil)
+		if err != nil || eng != nil {
+			t.Fatalf("buildBurstProcessor = (%v, %v), want (nil, nil)", eng, err)
+		}
+	})
+	t.Run("onboard model equals primary", func(t *testing.T) {
+		base.OnboardSummarizerModel = "koala/phi4-mini"
+		eng, err := buildBurstProcessor(base, nil)
+		if err != nil || eng != nil {
+			t.Fatalf("buildBurstProcessor = (%v, %v), want (nil, nil)", eng, err)
+		}
+	})
+}

cmd/tapir/report.go

@@ -6,6 +6,7 @@ import (
 	"io"
 	"os"
 	"text/tabwriter"
+	"time"
 
 	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
 )
@@ -14,6 +15,27 @@ import (
 // weeks. The gate passes when any user reaches it.
 const gateThreshold = 2
 
+// defaultGateStart is the date Stage-0 return-usage tracking begins: the morning
+// the pilot was actually unblocked and summaries started flowing (2026-06-11).
+// Activity before this — testing, the period the pilot was stuck on zero — is
+// noise and must not count toward the gate. Override with TAPIR_USAGE_GATE_START
+// (YYYY-MM-DD). The gate measures whether users RETURN once it genuinely works.
+const defaultGateStart = "2026-06-11"
+
+// gateStart resolves the baseline date from TAPIR_USAGE_GATE_START or the default,
+// parsed as a UTC calendar day.
+func gateStart() (time.Time, error) {
+	v := os.Getenv("TAPIR_USAGE_GATE_START")
+	if v == "" {
+		v = defaultGateStart
+	}
+	t, err := time.Parse("2006-01-02", v)
+	if err != nil {
+		return time.Time{}, fmt.Errorf("TAPIR_USAGE_GATE_START=%q: want YYYY-MM-DD: %w", v, err)
+	}
+	return t, nil
+}
+
 // 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).
@@ -28,16 +50,24 @@ func runReport(ctx context.Context, _ []string) error {
 	}
 	defer s.Close()
 
-	rows, err := s.ActiveWeeks(ctx)
+	since, err := gateStart()
 	if err != nil {
 		return err
 	}
-	return formatReport(os.Stdout, rows)
+
+	rows, err := s.ActiveWeeks(ctx, since)
+	if err != nil {
+		return err
+	}
+	return formatReport(os.Stdout, rows, since)
 }
 
 // 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 {
+func formatReport(w io.Writer, rows []store.UserActiveWeeks, since time.Time) error {
+	if _, err := fmt.Fprintf(w, "Counting usage since %s (Stage-0 gate baseline)\n\n", since.Format("2006-01-02")); err != nil {
+		return err
+	}
 	if len(rows) == 0 {
 		_, err := fmt.Fprintln(w, "no users yet")
 		return err

cmd/tapir/report_test.go

@@ -3,12 +3,15 @@ package main
 import (
 	"strings"
 	"testing"
+	"time"
 
 	"github.com/stretchr/testify/require"
 
 	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
 )
 
+var testSince = time.Date(2026, 6, 11, 0, 0, 0, 0, time.UTC)
+
 func TestFormatReportColumnsAndGatePass(t *testing.T) {
 	rows := []store.UserActiveWeeks{
 		{UserID: "user-a", DisplayName: "Ada", ActiveWeeks: 3},
@@ -16,9 +19,10 @@ func TestFormatReportColumnsAndGatePass(t *testing.T) {
 	}
 
 	var b strings.Builder
-	require.NoError(t, formatReport(&b, rows))
+	require.NoError(t, formatReport(&b, rows, testSince))
 	out := b.String()
 
+	require.Contains(t, out, "since 2026-06-11", "report states the gate baseline date")
 	require.Contains(t, out, "USER")
 	require.Contains(t, out, "ACTIVE_WEEKS")
 	require.Contains(t, out, "Ada")
@@ -34,12 +38,12 @@ func TestFormatReportGateNotMet(t *testing.T) {
 	rows := []store.UserActiveWeeks{{UserID: "user-a", ActiveWeeks: 1}}
 
 	var b strings.Builder
-	require.NoError(t, formatReport(&b, rows))
+	require.NoError(t, formatReport(&b, rows, testSince))
 	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.NoError(t, formatReport(&b, nil, testSince))
 	require.Contains(t, b.String(), "no users yet")
 }

cmd/tapir/scheduler.go

@@ -143,8 +143,18 @@ func runScheduler(
 		return // disabled
 	}
 
-	pass := 0
+	// Derive the rotation offset from wall-clock, NOT an in-memory counter. A
-	runDiscoveryPass(ctx, pass, lister, runUser, log)
+	// counter reset to 0 on every pod restart always hands the lead to the
+	// first-listed user — so frequent deploys re-starve whoever is last (exactly
+	// what happened to the first pilot user during a deploy-heavy session). A
+	// time-based offset advances with real time and is identical across restarts,
+	// so the lead rotates fairly regardless of how often the pod bounces.
+	runPass := func() {
+		pass := int(time.Now().Unix() / int64(interval/time.Second))
+		runDiscoveryPass(ctx, pass, lister, runUser, log)
+	}
+
+	runPass()
 
 	ticker := time.NewTicker(interval)
 	defer ticker.Stop()
@@ -153,8 +163,7 @@ func runScheduler(
 		case <-ctx.Done():
 			return
 		case <-ticker.C:
-			pass++
+			runPass()
-			runDiscoveryPass(ctx, pass, lister, runUser, log)
 		}
 	}
 }

docs/architecture/architecture.md

@@ -250,6 +250,32 @@ 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.
 
+### Connect-time onboarding burst (ADR-018 → ADR-028)
+
+On a successful YouTube connect, `ConnectHandler` enqueues a connect-triggered discovery pass;
+the `discoveryTrigger` runs that pass and then fires the **onboarding burst** — a third entry path
+that summarises up to `TAPIR_ONBOARD_SUMMARIZE_COUNT` (default 3, hard-capped) of the new user's
+videos so the first session is not empty. The burst still flows through `globalFetchGate` (it is
+not a throughput change); ADR-028 sharpened *which* videos and *which model*:
+
+- **Selection** is `OnboardBurstVideoIDs`, not pure newest-first. It keeps newest-first order but
+  excludes a video whose **known** duration is outside `[TAPIR_MIN_VIDEO_SECONDS,
+  TAPIR_ONBOARD_MAX_VIDEO_SECONDS]` (drops Shorts and multi-hour livestream VODs). An unknown
+  (NULL) duration is degrade-open — kept, but ranked after known-good rows. The connect-triggered
+  discovery pass runs *before* the burst, and ADR-023's `videos.list` enrichment now **persists**
+  `duration_s` (instead of discarding it after the Shorts filter), so a fresh user's candidates
+  carry a duration in time for selection.
+- **Model**: the burst runs through a dedicated summarizer chain led by
+  `TAPIR_ONBOARD_SUMMARIZER_MODEL` (default `iguana/gemma4-26b`, the stronger local model), with
+  the standard ADR-022 chain following as fallback. This is a wiring choice — a second
+  `engineProcessor` over the same store / transcript cache / sink; the engine and ports are
+  unchanged. Empty / equal-to-primary collapses it back onto the shared processor.
+
+`has-captions` is deliberately **not** a selection signal — it is only knowable after a gate fetch
+(or a ~0-probability cache hit at pilot scale), so the burst can avoid known-junk but cannot
+promise captions. Cached-transcript-first selection was investigated and rejected (ADR-028:
+~3% cross-user overlap).
+
 ---
 
 ## Sequence — core use case: new video summarized

docs/homelab-integration.md

@@ -30,7 +30,9 @@ it** — endpoints and aliases drift, and this file is a snapshot (2026-06-06),
 - **Summarizer fallback chain (ADR-022).** The primary alias is the *first* of an ordered chain;
   on failure or unparseable output the summarizer advances to the next model. All reached through
   the same gateway by alias.
-  - `TAPIR_FALLBACK_MODEL` — local fallback. **Default `koala/phi4-14b`.** Empty disables it.
+  - `TAPIR_FALLBACK_MODEL` — local fallback. **Default `iguana/gemma4-26b`** — on iguana, NOT
+    koala, so the fallback does not compete with koala's other GPU loads (and runs from a different
+    egress IP). Empty disables it.
   - `TAPIR_CLOUD_FALLBACK_MODEL` — worst-case EXTERNAL fallback. **Default `berget/mistral-small`.**
     **Set this empty (`""`) for any client/NDA deployment** so content never leaves the local
     stack — the chain then contains only local endpoints.
@@ -222,6 +224,9 @@ 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_USAGE_GATE_START` — `YYYY-MM-DD`, default **`2026-06-11`** (the morning the pilot was
+  unblocked and summaries started flowing). `tapir report` counts return-usage (distinct active
+  weeks, ADR-016) only from this date, so pre-launch testing and the blocked period are excluded.
 - `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`

docs/specs/chat-with-transcript.md

@@ -0,0 +1,89 @@
+# Spec — Chat with a video's stored transcript (ADR-027)
+
+**Repo:** tapir · **Size:** medium · **Solo session.** Implements ADR-027. Read CLAUDE.md,
+DECISIONS.md (ADR-021 transcript store, ADR-022 model chain, ADR-012 isolation, ADR-027), and
+`docs/ui-spec.md` first. TBD, conventional commits, `task check` green per commit, `templ
+generate` after view changes.
+
+## FIRST: append ADR-027 to DECISIONS.md
+ADR-027 text is provided separately (planning thread). Insert immediately before the
+`## Rejected alternatives` heading, as the first commit, so the decision precedes the build.
+
+## What this is
+A per-video chat letting the user ask questions against a video's **already-stored** transcript,
+entered from the summary view. Born from observed demand: the maintainer read real summaries and
+some made him want to dig deeper — this gives that "I want more" reaction somewhere to go, without
+watching the video.
+
+## HARD CONSTRAINT — stored-transcript-only (the safety property)
+Chat is available **ONLY** for videos that already have a stored transcript (ADR-021). It must
+**never** trigger a caption fetch, never touch the rate gate, never reach YouTube. Entry being
+"from a summarized video" guarantees the transcript exists. If somehow invoked on a video with no
+stored transcript → show "transcript not available for chat", NO fetch. This is what makes the
+feature safe by construction; do not add an on-demand-fetch path (explicitly deferred).
+
+## 1. Entry point
+- A "Dig deeper" / "Ask about this" affordance on the **summary view** of a summarized video
+  (not the list cards — the detail/summary page). Quiet, consistent with the existing card-state
+  styling.
+- Opens a chat panel/view scoped to that one video, with its stored transcript as context.
+
+## 2. The chat
+- Read the stored transcript for the video (via the ADR-021 `TranscriptStore`, keyed by
+  `(provider, provider_video_id)`). No fetch.
+- Send transcript + the user's question + minimal system framing to the chosen model via the
+  **existing LiteLLM gateway** (the same client the summarizer uses — a chat is a different
+  call, not a new integration).
+- Stream or return the answer; render in the chat panel. HTMX/no-JS ethos — match the existing
+  app (the summarize status uses HTMX polling; chat can use a simple POST-and-render or HTMX
+  streaming if clean).
+- **Transcript truncation:** reuse/respect `TAPIR_MAX_TRANSCRIPT_CHARS` (ADR-022) so a long
+  transcript fits the model context. If truncated, the chat should be honest that it's working
+  from a bounded portion (a quiet note), since answers about the tail of a long video may be
+  incomplete.
+
+## 3. Model selection (the instrumentation win)
+- **Default model = the model that produced this video's summary.** (Store/lookup which chain
+  model summarized it — if not already recorded, this is a small addition; if recording it is
+  non-trivial, default to the chain primary and note the gap.)
+- **User can switch** among the ADR-022 chain models (`phi4-mini`, `gemma4-26b`,
+  `mistral-small` to start) via a simple selector in the chat panel. Switching re-runs against
+  the same transcript — this is deliberate model-comparison instrumentation.
+- Respect the local-first / NDA posture: if `TAPIR_CLOUD_FALLBACK_MODEL=""` (cloud disabled),
+  the external model is NOT offered in the switcher — only local models. Chat must honor the same
+  "content stays local" guarantee as ADR-022.
+
+## 4. Ephemeral (v1)
+- No persisted chat history. Conversation lives for the session/page. No new table, no migration.
+- (Multi-turn within a session is fine — keep the running messages in the request/page state —
+  but nothing is written to the DB.)
+
+## 5. Isolation
+- The transcript is shared/non-RLS (ADR-021) — fine, it's public content. But the chat is invoked
+  by a user about a video **in their feed**; confirm the entry path is reachable only for the
+  requesting user's own videos (the summary view is already RLS-scoped). Chat adds no new
+  user-data surface (ephemeral), so there's nothing new to RLS — but the test should confirm a
+  user can only open chat from their own summary view, not arbitrary video ids.
+
+## Tests
+- Chat on a video with a stored transcript → answer returned; assert NO caption-fetch / no
+  YouTube call occurs (the safety property — this is the key assertion).
+- Chat invoked on a video with no stored transcript → honest "not available", NO fetch.
+- Model switch → re-runs against the same transcript with the selected model; cloud model absent
+  from the switcher when `TAPIR_CLOUD_FALLBACK_MODEL=""`.
+- Truncation honored for a long transcript; the bounded-context note shows.
+- Entry is reachable only from the user's own summary view (isolation).
+
+## Out of scope / deferred (record, don't build)
+- **Persisted chat history** (per-user, RLS-scoped) — deferred until evidence anyone revisits a
+  conversation.
+- **Show-source / transcript-verification UI** — the natural v2 (ADR-027 records it); v1 is
+  chat-only/trust-the-model. ADR-021's stored transcript makes v2 cheap when wanted.
+- **On-demand fetch** for un-stored videos — would reintroduce the caption-fetch surface the
+  stored-only constraint removes. Not now.
+- Anything that nudges the user to return (ADR-020 — gate contamination).
+
+## Boundaries
+Stored-transcript-only (HARD). No rate-gate/fetch surface. No auth changes. No new persisted
+state in v1. Reuse the existing gateway client + truncation config; don't build a new model
+integration.

docs/specs/onboarding-wow-burst.md

@@ -0,0 +1,128 @@
+# Spec — Onboarding "wow" burst: better picks, stronger model
+
+**Repo:** tapir · **Size:** medium · **Solo session** (not a swarm).
+
+> **Status: built (v0.25.0, ADR-028).** This supersedes the original investigate-first brief
+> (committed as the prior version of this file): Phase 1 was run against the live pilot DB and its
+> findings are folded into "Why this exists" below; Phase 2 was built as described here. The one
+> brief lever NOT built — the honest "the rest fill in over the coming days" framing copy — is
+> listed under *Explicitly NOT in this slice*.
+
+**Why this exists.** A new user's first session decides whether they return (the Stage-0 gate,
+VISION.md). On connect, Tapir fires a capped burst (≤`TAPIR_ONBOARD_SUMMARIZE_COUNT`, default 3)
+that summarizes the user's newest unsummarized videos so the feed isn't empty (the burst itself
+works — wired in `cmd/tapir/discovery.go` → `cmd/tapir/main.go` `onboard`). A Phase-1
+investigation of the live pilot DB found the burst *fires* but delivers a **weak first
+impression** for two concrete reasons, and ruled out a third idea:
+
+1. **Picks are junk.** Selection is pure newest-first (`videos.NewestUnsummarizedVideoIDs`,
+   `ORDER BY published_at DESC`) with **zero quality signal**. Pilot user "Jonte"'s live burst-3
+   were a stock-ticker **livestream** + two regional news clips — the newest, not the best.
+2. **Weakest model on the first impression.** All of Jonte's summaries ran on
+   `koala/phi4-mini` (the documented weak link — ADR-022 was born from its failures). The
+   stronger, brain-validated `iguana/gemma4-26b` was never used for the burst.
+3. **Cached-first is empty at pilot scale — REJECTED.** The idea (summarize already-cached
+   transcripts instantly, zero fetch) dies on the numbers: only **11 videos** overlap between the
+   two pilot users (~3% of each library), **0** cached-and-unsummarized, and a new user's
+   newest-20 unsummarized are **20/20 NOT cached** — newest-first and cached-first are
+   structurally incompatible (fresh uploads are exactly what nobody has fetched yet). Not built.
+
+This is a **curation/latency problem for ~3 videos, NOT a throughput/429 problem** — fetching 3
+captions is nowhere near the rate limit. Nothing here fetches harder or pressures the rate gate;
+it picks the right few videos and runs a better model on them.
+
+Read `CLAUDE.md`, `DECISIONS.md` (esp. ADR-014, ADR-018, ADR-020, ADR-021, ADR-022, ADR-023,
+and the new **ADR-028**), and `VISION.md` (the Stage-0 gate) 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 (none expected).
+
+## Decisions already made (do not reopen)
+- **Not a throughput change.** The caption rate gate (ADR-014) is untouched — same pacing, same
+  priority lane (ADR-026). This slice changes *which* ≤3 videos the burst spends its fetches on
+  and *which model* summarizes them, never how fast or how many.
+- **Cached-first is dropped** (ADR-028, the 3% overlap). The engine's existing read-stored-first
+  (ADR-021, `resolveTranscript`) stays — it already gives a free instant summary on the rare
+  cache hit, transparently. We do not *select* for cache hits.
+- **has-captions is not a pre-fetch signal.** It is only knowable after a gate fetch (or a cache
+  hit, ~0 for new videos). Selection can only *avoid known-junk* (Shorts/live/over-long) — it
+  cannot *guarantee* captions. The spec is honest about this: better odds, not a promise.
+- **No credentialed caption fetch** (ADR-010/ADR-026 dead end). **No client extension.**
+
+## 1. Persist `duration_s` at discovery (the enabling change)
+The `videos.duration_s` column exists (migration 001) but is **never written** — ADR-023's
+`filterLowValue` (`internal/adapters/youtube/youtube.go`) already fetches each candidate's
+duration via the cheap quota `videos.list` call, uses it to drop Shorts/live, then **discards
+it**. Stop discarding:
+- Add `DurationSeconds int` to `domain.Video`.
+- In `filterLowValue`, set `DurationSeconds` on each kept video from the `videos.list` `meta`.
+- `UpsertVideo` writes `duration_s`, **COALESCE-preserving** a known value (never overwrite a
+  real duration with 0/unknown), mirroring the `channel_title` backfill stance (migration 014).
+- No new migration — the column is already there.
+
+Consequence: a fresh user's connect-triggered discovery pass runs **before** the onboard burst
+(`Enqueue`: `run()` then `onboard()`), so duration is populated for the burst's candidates at
+connect. Existing rows backfill on their next discovery pass; until then their `duration_s` is
+NULL and treated as "unknown" (§2).
+
+## 2. Junk-avoiding burst selection
+New store method, RLS-scoped via `withUser`:
+
+```
+OnboardBurstVideoIDs(ctx, userID string, limit, minSeconds, maxSeconds int) ([]string, error)
+```
+
+- Same base as the old `NewestUnsummarizedVideoIDs`: the user's videos with no summary yet,
+  `ORDER BY published_at DESC NULLS LAST, seen_at DESC`, `LIMIT limit`.
+- **Exclude known-junk**: a row is dropped only when `duration_s IS NOT NULL` **and**
+  (`duration_s < minSeconds` OR `duration_s > maxSeconds`). A NULL duration is **unknown** — kept
+  (degrade-open: never starve the burst because metadata is missing), but ordered *after* rows
+  with a known-good duration so a freshly-enriched good pick wins when both exist.
+- `minSeconds` reuses `TAPIR_MIN_VIDEO_SECONDS` (default 60 — the Shorts floor, ADR-023).
+  `maxSeconds` is new: `TAPIR_ONBOARD_MAX_VIDEO_SECONDS` (default 14400 = 4h) — drops the
+  multi-hour livestream VODs that pass the live filter once ended.
+- `minSeconds<=0` and `maxSeconds<=0` each disable that bound (so `0/0` == the old
+  newest-first behaviour, the reversibility lever).
+- The burst switches to this method; `NewestUnsummarizedVideoIDs` is removed (fully superseded —
+  `OnboardBurstVideoIDs(., 0, 0)` is identical pure-newest behaviour).
+
+## 3. Stronger model for the burst
+The burst summarizes only ≤3 videos, so a slower, stronger model is affordable exactly here.
+- New config `TAPIR_ONBOARD_SUMMARIZER_MODEL` (default `iguana/gemma4-26b` — the brain-validated
+  homelab general-purpose model, already the ADR-022 fallback).
+- Build a **burst-specific summarizer chain** that puts the onboard model **first**, then the
+  standard chain (primary → local fallback → cloud) as resilience, deduped. Wrap it in a
+  burst-specific `engineProcessor` reusing the same store/transcript-cache/sink — a pure wiring
+  choice, engine and ports unchanged (Clean Architecture, ADR-003).
+- The `onboard` closure uses the burst processor instead of `app.Processor`.
+- **Collapse cleanly**: when `OnboardSummarizerModel` is empty or equals `SummarizerModel`, the
+  onboard path reuses `app.Processor` (no separate chain) — the reversibility lever.
+- Local-first preserved: the onboard model is a local alias; the cloud endpoint stays last in the
+  chain, so a client/NDA deployment with `TAPIR_CLOUD_FALLBACK_MODEL=""` keeps burst content
+  local too.
+
+## 4. Behaviour spec + docs
+- Add scenarios to `docs/use-cases/connect_account.feature` (the connect → burst flow): burst
+  skips a too-long/live video in favour of a reasonable-length one; burst summarizes with the
+  stronger model first. Map them in `scenarioCoverage` so `TestScenarioCoverage` stays green.
+- Update `docs/architecture/architecture.md` (the onboarding-burst section) to describe the
+  junk-avoiding selection + the burst model override.
+- ADR-028 in `DECISIONS.md` records the rationale (incl. the rejected cached-first lever).
+
+## Success criteria
+- `task check` green (fmt, vet, lint, `go test -p 1 ./...`).
+- A unit test proves `OnboardBurstVideoIDs` drops a known too-long / sub-min video and keeps a
+  good one, newest-first, RLS-scoped, unsummarized-only.
+- A test proves discovery persists `duration_s` and does not clobber it on re-upsert.
+- A test proves the burst chain leads with the onboard model (then the standard chain).
+- Config defaults + bounds tested (`OnboardMaxVideoSeconds`, `OnboardSummarizerModel`).
+- No change to the rate gate, fetch pacing, or burst cap. `0/0` + empty model == prior behaviour.
+
+## Explicitly NOT in this slice
+- Cached-first selection (rejected, ADR-028).
+- Any caption-availability *guarantee* (impossible pre-fetch).
+- **Honest "taster" framing copy** ("summaries of a few of your videos to get you started — the
+  rest fill in over the coming days"). A good lever from the original brief, but it's a UI/copy
+  change with no backend dependency; deferred to a UI pass, tracked as an issue.
+- Backfilling `duration_s` for existing rows via a migration (it backfills lazily on discovery).
+- Return-nudges / digests (ADR-020: poisons the unprompted-return signal).
+- Raising fetch throughput, multi-IP, or Whisper (out of scope; the gate is deliberate).

docs/use-cases/chat_transcript.feature

@@ -0,0 +1,69 @@
+Feature: Chat with a video's stored transcript
+  As a reader whose summary made me want to dig deeper
+  I want to ask questions about the video without watching it
+  So that I can go further on the ones worth it, without leaving the reader
+
+  # ADR-027. The load-bearing constraint is safety-by-construction: chat runs
+  # ONLY against an already-stored transcript (ADR-021) and never fetches captions,
+  # never touches the rate gate, never reaches YouTube. Entry is from the summary
+  # view of one's OWN video; the conversation is ephemeral (no persisted history).
+
+  Background:
+    Given I have a summarized video with a stored transcript
+
+  Scenario: A summary view offers a deeper-dive into the video
+    When I view the summary
+    Then I see a "dig deeper" affordance that opens a chat about this video
+    And it opens the chat in place, below the summary, without leaving the page
+
+  Scenario: The summary and the chat are on one page
+    When I open the chat
+    Then the summary stays visible alongside the chat
+    And I can read the summary while I ask questions
+
+  Scenario: Ask a question answered from the stored transcript
+    When I ask a question in the chat
+    Then the answer is produced from the stored transcript
+    And no caption fetch and no YouTube call occurs
+
+  Scenario: Chat never fetches captions or reaches YouTube
+    When I ask a question in the chat
+    Then Tapir reads only the stored transcript
+    And the caption-fetch and video-fetch paths are never invoked
+
+  Scenario: A video with no stored transcript offers no chat
+    Given a video that has no stored transcript
+    When I open the chat for it
+    Then I am told chat is not available
+    And no fetch is attempted and no model is called
+
+  Scenario: The default model is the summary's model and is switchable
+    When I open the chat
+    Then the model defaults to the model that produced the summary
+    And I can switch among the offered chain models
+
+  Scenario: Switching models re-runs against the same transcript
+    When I ask a question with a different chain model selected
+    Then the chosen model answers
+    And it answers against the same stored transcript
+
+  Scenario: The cloud model is hidden when cloud is disabled
+    Given the cloud fallback model is disabled
+    Then the chat switcher offers only local models
+
+  Scenario: A long transcript is bounded and the chat says so
+    Given the stored transcript is longer than the model budget
+    When I ask a question
+    Then the answer is produced from a bounded portion
+    And the chat notes that it worked from a bounded portion
+
+  Scenario: A multi-turn conversation is ephemeral
+    When I ask a follow-up question
+    Then the prior turn is carried into the answer
+    And nothing about the conversation is written to the database
+
+  Scenario: Chat is reachable only from my own summary view
+    Given another user has a summarized video with a stored transcript
+    When I try to open the chat for their video
+    Then I get a not-found response
+    And no model is called

docs/use-cases/connect_account.feature

@@ -24,12 +24,19 @@ Feature: Connect and manage video accounts
     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: Connecting summarizes my newest videos right away
+  Scenario: Connecting summarizes my best recent videos right away
     Given I have no connected video accounts
     When I connect my YouTube account
-    Then up to the onboarding cap of my newest videos are summarized through the rate gate
+    Then up to the onboarding cap of my newest likely-good videos are summarized through the rate gate
+    And videos whose known duration is too short or too long are skipped
     And the rest are left to the scheduled recency-bounded pass
 
+  Scenario: The onboarding burst summarizes with a stronger model
+    Given I have no connected video accounts
+    When I connect my YouTube account
+    Then the burst summarizes with the stronger onboarding model first
+    And the standard summarizer chain still follows as a fallback
+
   Scenario: Tokens are never stored in the clear
     When I connect any video account
     Then no OAuth token value is stored in the database

internal/adapters/chat/chat.go

@@ -0,0 +1,171 @@
+// Package chat implements the per-video deeper-dive chat (ADR-027): a read-only
+// QA over a video's ALREADY-STORED transcript (ADR-021). It is the enforcement
+// point for the feature's load-bearing safety property — stored-transcript-only:
+// the Service has NO VideoSource and NO caption-fetch dependency, only a
+// Completer factory, so it CANNOT reach YouTube or the rate gate by construction.
+// The caller supplies the stored transcript text; chat never fetches.
+//
+// It reuses the same LiteLLM gateway as the summarizer (a chat is a different
+// call, not a new integration) and the same transcript-truncation discipline
+// (TAPIR_MAX_TRANSCRIPT_CHARS) so a long transcript fits a small-context model.
+package chat
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"unicode/utf8"
+)
+
+// Completer is the minimal LLM chat surface the Service needs. *llm.Client
+// satisfies it; tests use a fake. It is the SAME surface the summarizer uses.
+type Completer interface {
+	Complete(ctx context.Context, system, user string) (string, error)
+}
+
+// Turn is one completed exchange in an ephemeral, session-only conversation
+// (ADR-027 v1: nothing is persisted).
+type Turn struct {
+	Question string
+	Answer   string
+}
+
+// Request is one chat turn: the chosen model, the stored transcript text, the
+// prior turns (for multi-turn context within the session), and the new question.
+type Request struct {
+	Model      string
+	Transcript string
+	History    []Turn
+	Question   string
+}
+
+// Reply is the model's answer plus whether the transcript was bounded to fit the
+// model context (so the UI can be honest that an answer about the tail of a long
+// video may be incomplete).
+type Reply struct {
+	Answer    string
+	Truncated bool
+}
+
+// Service answers questions against a stored transcript via a switchable set of
+// models. models is the ordered, local-first list offered to the user (the cloud
+// model is simply absent when disabled — see cmd wiring); maxChars bounds the
+// transcript sent to any model (0 = unbounded). newClient builds a Completer for
+// a chosen model alias (the same gateway, a different alias).
+type Service struct {
+	newClient func(model string) Completer
+	models    []string
+	maxChars  int
+}
+
+// New constructs a Service. models must be non-empty and already filtered to the
+// offerable set (cloud excluded when disabled) and de-duplicated by the caller.
+func New(newClient func(model string) Completer, models []string, maxChars int) *Service {
+	return &Service{newClient: newClient, models: models, maxChars: maxChars}
+}
+
+// Models returns a copy of the offerable model list (local-first order).
+func (s *Service) Models() []string {
+	out := make([]string, len(s.models))
+	copy(out, s.models)
+	return out
+}
+
+// offers reports whether model is in the offerable set — the guard that keeps an
+// arbitrary, un-offered alias (e.g. a forged form value) from reaching the gateway.
+func (s *Service) offers(model string) bool {
+	for _, m := range s.models {
+		if m == model {
+			return true
+		}
+	}
+	return false
+}
+
+// DefaultModel resolves the model a fresh chat opens with: the summary's own
+// model when it is still an offered option (the ADR-027 default — chat continues
+// in the model that produced the summary), otherwise the first offered model.
+// Returns "" only when no models are configured.
+func (s *Service) DefaultModel(summaryModel string) string {
+	if summaryModel != "" && s.offers(summaryModel) {
+		return summaryModel
+	}
+	if len(s.models) > 0 {
+		return s.models[0]
+	}
+	return ""
+}
+
+// Answer runs one chat turn. The model is forced back to a default if the request
+// names an un-offered alias, so chat can never call the gateway with an arbitrary
+// model. The transcript is truncated up front (reporting whether it was cut) and
+// passed as system context; the running conversation is the user message.
+func (s *Service) Answer(ctx context.Context, req Request) (Reply, error) {
+	if len(s.models) == 0 {
+		return Reply{}, fmt.Errorf("chat: no models configured")
+	}
+	model := req.Model
+	if !s.offers(model) {
+		model = s.DefaultModel("")
+	}
+
+	transcript, truncated := truncate(req.Transcript, s.maxChars)
+	system := buildSystem(transcript, truncated)
+	user := buildUser(req.History, req.Question)
+
+	out, err := s.newClient(model).Complete(ctx, system, user)
+	if err != nil {
+		return Reply{}, fmt.Errorf("chat: %s: %w", model, err)
+	}
+	answer := strings.TrimSpace(out)
+	if answer == "" {
+		return Reply{}, fmt.Errorf("chat: %s returned an empty answer", model)
+	}
+	return Reply{Answer: answer, Truncated: truncated}, nil
+}
+
+const systemPreamble = `You are Tapir, answering questions about ONE video using ONLY the transcript below.
+Ground every answer in the transcript. If the transcript does not contain the answer, say so plainly rather than guessing.`
+
+const truncatedNote = `
+The transcript below is truncated to fit the model — if a question seems to concern something missing, note it may be beyond the available portion.`
+
+// buildSystem frames the model as a transcript-grounded QA assistant and embeds
+// the (possibly truncated) transcript as context.
+func buildSystem(transcript string, truncated bool) string {
+	var b strings.Builder
+	b.WriteString(systemPreamble)
+	if truncated {
+		b.WriteString(truncatedNote)
+	}
+	b.WriteString("\n\nTranscript:\n")
+	b.WriteString(transcript)
+	return b.String()
+}
+
+// buildUser renders the running conversation as the user message: prior turns as
+// Q/A pairs followed by the new question. Folding history into one message keeps
+// the Completer surface (a single system+user call) unchanged — no new llm method.
+func buildUser(history []Turn, question string) string {
+	var b strings.Builder
+	for _, t := range history {
+		fmt.Fprintf(&b, "Q: %s\nA: %s\n\n", t.Question, t.Answer)
+	}
+	fmt.Fprintf(&b, "Q: %s", question)
+	return b.String()
+}
+
+// truncate caps content to max bytes on a UTF-8 rune boundary, reporting whether
+// it cut. It mirrors the summarizer's truncation discipline (ADR-022) but returns
+// the cut flag so the chat UI can be honest about a bounded transcript. A
+// non-positive max (or content already within budget) returns content unchanged.
+func truncate(content string, max int) (string, bool) {
+	if max <= 0 || len(content) <= max {
+		return content, false
+	}
+	cut := max
+	for cut > 0 && !utf8.RuneStart(content[cut]) {
+		cut--
+	}
+	return content[:cut], true
+}

internal/adapters/chat/chat_test.go

@@ -0,0 +1,164 @@
+package chat
+
+import (
+	"context"
+	"errors"
+	"strings"
+	"testing"
+)
+
+// recordingCompleter captures the system+user it was asked with and returns a
+// canned answer (or error). It also records which model alias built it.
+type recordingCompleter struct {
+	model      string
+	lastSystem string
+	lastUser   string
+	answer     string
+	err        error
+	calls      *int
+}
+
+func (c *recordingCompleter) Complete(_ context.Context, system, user string) (string, error) {
+	*c.calls++
+	c.lastSystem = system
+	c.lastUser = user
+	if c.err != nil {
+		return "", c.err
+	}
+	return c.answer, nil
+}
+
+// factory builds a recordingCompleter per model and records the last one built so
+// the test can assert which model alias was actually used for the gateway call.
+type factory struct {
+	answer string
+	err    error
+	calls  int
+	used   *recordingCompleter
+}
+
+func (f *factory) make(model string) Completer {
+	c := &recordingCompleter{model: model, answer: f.answer, err: f.err, calls: &f.calls}
+	f.used = c
+	return c
+}
+
+func TestModelsAreOfferedLocalFirstAndCopied(t *testing.T) {
+	f := &factory{answer: "ok"}
+	s := New(f.make, []string{"koala/phi4-mini", "iguana/gemma4-26b"}, 0)
+
+	got := s.Models()
+	want := []string{"koala/phi4-mini", "iguana/gemma4-26b"}
+	if len(got) != len(want) || got[0] != want[0] || got[1] != want[1] {
+		t.Fatalf("Models() = %v, want %v", got, want)
+	}
+	// Mutating the returned slice must not corrupt the Service's list.
+	got[0] = "tampered"
+	if s.Models()[0] != "koala/phi4-mini" {
+		t.Fatal("Models() leaked its backing slice")
+	}
+}
+
+func TestDefaultModelIsTheSummarysModelWhenOffered(t *testing.T) {
+	f := &factory{answer: "ok"}
+	s := New(f.make, []string{"koala/phi4-mini", "iguana/gemma4-26b", "berget/mistral-small"}, 0)
+
+	if got := s.DefaultModel("iguana/gemma4-26b"); got != "iguana/gemma4-26b" {
+		t.Fatalf("DefaultModel(summary) = %q, want the summary's model", got)
+	}
+	// A summary model no longer offered (e.g. cloud disabled) falls back to first.
+	if got := s.DefaultModel("berget/old-model"); got != "koala/phi4-mini" {
+		t.Fatalf("DefaultModel(un-offered) = %q, want the first offered model", got)
+	}
+	// No summary model recorded → first offered.
+	if got := s.DefaultModel(""); got != "koala/phi4-mini" {
+		t.Fatalf("DefaultModel(\"\") = %q, want the first offered model", got)
+	}
+}
+
+func TestAnswerGroundsOnTranscriptAndCarriesHistory(t *testing.T) {
+	f := &factory{answer: "  The video is about attention.  "}
+	s := New(f.make, []string{"koala/phi4-mini"}, 0)
+
+	reply, err := s.Answer(context.Background(), Request{
+		Model:      "koala/phi4-mini",
+		Transcript: "ATTENTION-TRANSCRIPT-MARKER",
+		History:    []Turn{{Question: "who", Answer: "the host"}},
+		Question:   "what is it about",
+	})
+	if err != nil {
+		t.Fatalf("Answer: %v", err)
+	}
+	if reply.Answer != "The video is about attention." {
+		t.Fatalf("answer not trimmed: %q", reply.Answer)
+	}
+	if reply.Truncated {
+		t.Fatal("short transcript must not report truncated")
+	}
+	// The transcript rides in the system prompt; the conversation in the user msg.
+	if !strings.Contains(f.used.lastSystem, "ATTENTION-TRANSCRIPT-MARKER") {
+		t.Fatal("transcript not grounded into the system prompt")
+	}
+	if !strings.Contains(f.used.lastUser, "Q: who") || !strings.Contains(f.used.lastUser, "A: the host") {
+		t.Fatalf("history not carried into the user message: %q", f.used.lastUser)
+	}
+	if !strings.Contains(f.used.lastUser, "what is it about") {
+		t.Fatal("new question missing from the user message")
+	}
+}
+
+func TestAnswerTruncatesLongTranscriptAndReportsIt(t *testing.T) {
+	f := &factory{answer: "answer"}
+	s := New(f.make, []string{"koala/phi4-mini"}, 10)
+
+	reply, err := s.Answer(context.Background(), Request{
+		Model:      "koala/phi4-mini",
+		Transcript: strings.Repeat("x", 500),
+		Question:   "summarize",
+	})
+	if err != nil {
+		t.Fatalf("Answer: %v", err)
+	}
+	if !reply.Truncated {
+		t.Fatal("a transcript past maxChars must report Truncated")
+	}
+	if strings.Count(f.used.lastSystem, "x") != 10 {
+		t.Fatalf("transcript not bounded to maxChars: got %d x's", strings.Count(f.used.lastSystem, "x"))
+	}
+}
+
+func TestAnswerForcesAnUnofferedModelBackToDefault(t *testing.T) {
+	f := &factory{answer: "answer"}
+	s := New(f.make, []string{"koala/phi4-mini", "iguana/gemma4-26b"}, 0)
+
+	// A forged/un-offered model must never reach the gateway as-is — it is forced
+	// to the default offered model (the cloud-absent guarantee depends on this).
+	_, err := s.Answer(context.Background(), Request{
+		Model:      "berget/secret-cloud-model",
+		Transcript: "t",
+		Question:   "q",
+	})
+	if err != nil {
+		t.Fatalf("Answer: %v", err)
+	}
+	if f.used.model != "koala/phi4-mini" {
+		t.Fatalf("un-offered model reached the gateway as %q, want the default", f.used.model)
+	}
+}
+
+func TestAnswerPropagatesCompleterError(t *testing.T) {
+	f := &factory{err: errors.New("gateway down")}
+	s := New(f.make, []string{"koala/phi4-mini"}, 0)
+
+	_, err := s.Answer(context.Background(), Request{Model: "koala/phi4-mini", Transcript: "t", Question: "q"})
+	if err == nil {
+		t.Fatal("expected the gateway error to propagate")
+	}
+}
+
+func TestAnswerRejectsEmptyModelSet(t *testing.T) {
+	s := New(func(string) Completer { return nil }, nil, 0)
+	if _, err := s.Answer(context.Background(), Request{Question: "q"}); err == nil {
+		t.Fatal("expected an error when no models are configured")
+	}
+}

internal/adapters/store/report.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"fmt"
 	"sort"
+	"time"
 
 	"github.com/jackc/pgx/v5"
 )
@@ -32,7 +33,12 @@ type UserActiveWeeks struct {
 // 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) {
+// ActiveWeeks counts each user's distinct active weeks from `since` onward. A zero
+// `since` means no lower bound (count all history). The Stage-0 gate baseline is
+// set by the caller (the report command) to the date real usage tracking began,
+// so pre-launch noise — testing, the period the pilot was blocked — does not count
+// toward the return-usage signal (ADR-016).
+func (s *Store) ActiveWeeks(ctx context.Context, since time.Time) ([]UserActiveWeeks, error) {
 	userIDs, err := s.identityUserIDs(ctx)
 	if err != nil {
 		return nil, err
@@ -40,7 +46,7 @@ func (s *Store) ActiveWeeks(ctx context.Context) ([]UserActiveWeeks, error) {
 
 	out := make([]UserActiveWeeks, 0, len(userIDs))
 	for _, uid := range userIDs {
-		row, err := s.activeWeeksFor(ctx, uid)
+		row, err := s.activeWeeksFor(ctx, uid, since)
 		if err != nil {
 			return nil, err
 		}
@@ -85,18 +91,18 @@ func (s *Store) identityUserIDs(ctx context.Context) ([]string, error) {
 // 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) {
+func (s *Store) activeWeeksFor(ctx context.Context, userID string, since time.Time) (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
+			       FROM login_events WHERE user_id = $1 AND seen_at >= $2
 			     UNION
 			     SELECT date_trunc('week', acted_at)
-			       FROM summary_actions WHERE user_id = $1
+			       FROM summary_actions WHERE user_id = $1 AND acted_at >= $2
 			 )
-			 SELECT count(DISTINCT wk) FROM weeks`, userID).Scan(&res.ActiveWeeks); err != nil {
+			 SELECT count(DISTINCT wk) FROM weeks`, userID, since).Scan(&res.ActiveWeeks); err != nil {
 			return fmt.Errorf("store: count active weeks: %w", err)
 		}
 		if err := tx.QueryRow(ctx,

internal/adapters/store/report_test.go

@@ -3,6 +3,7 @@ package store_test
 import (
 	"context"
 	"testing"
+	"time"
 
 	"github.com/jackc/pgx/v5/pgxpool"
 	"github.com/stretchr/testify/require"
@@ -54,7 +55,7 @@ func TestActiveWeeksCountsDistinctWeeksAcrossReadsAndActs(t *testing.T) {
 		   ($1, 'vid-2', 'saved',   '2026-01-19T18:00:00Z')`, userA)
 	require.NoError(t, err)
 
-	got, err := s.ActiveWeeks(ctx)
+	got, err := s.ActiveWeeks(ctx, time.Time{}) // zero since = no lower bound
 	require.NoError(t, err)
 	require.Len(t, got, 2, "both identity users must appear")
 
@@ -72,7 +73,33 @@ func TestActiveWeeksEmptyWhenNoUsers(t *testing.T) {
 	s := newStore(t)
 	resetDB(t, rawPool(t))
 
-	got, err := s.ActiveWeeks(ctx)
+	got, err := s.ActiveWeeks(ctx, time.Time{})
 	require.NoError(t, err)
 	require.Empty(t, got)
 }
+
+// TestActiveWeeksExcludesBeforeGateStart proves the baseline cutoff: activity
+// before `since` does not count, so pre-launch noise (testing, the pilot's blocked
+// period) is excluded from the Stage-0 return-usage gate (ADR-016).
+func TestActiveWeeksExcludesBeforeGateStart(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	p := rawPool(t)
+	resetDB(t, p)
+
+	seedReportUser(t, p, userA, "subject-a", "Ada")
+
+	// One read well before the baseline, two reads in distinct weeks after it.
+	_, err := p.Exec(ctx,
+		`INSERT INTO login_events (user_id, seen_at) VALUES
+		   ($1, '2026-05-01T09:00:00Z'),
+		   ($1, '2026-06-12T09:00:00Z'),
+		   ($1, '2026-06-19T09:00:00Z')`, userA)
+	require.NoError(t, err)
+
+	since := time.Date(2026, 6, 11, 0, 0, 0, 0, time.UTC)
+	got, err := s.ActiveWeeks(ctx, since)
+	require.NoError(t, err)
+	require.Len(t, got, 1)
+	require.Equal(t, 2, got[0].ActiveWeeks, "only the two post-baseline weeks count; the May read is excluded")
+}

internal/adapters/store/store_test.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"fmt"
 	"os"
+	"path/filepath"
 	"testing"
 
 	embeddedpostgres "github.com/fergusstrange/embedded-postgres"
@@ -24,11 +25,22 @@ var _ ports.Sink = (*store.Store)(nil)
 var dsn string
 
 func TestMain(m *testing.M) {
-	const port = 54329
+	// Port + runtime/data dirs are per-process (PID-derived) so two concurrent
+	// `go test` invocations — e.g. a push-run and a tag-run firing together in CI —
+	// don't collide on a fixed port or a shared data dir (which silently failed
+	// both runs). CachePath is shared so the PG archive is downloaded once, not
+	// per process. Base 54000 keeps this package's range distinct from web's.
+	port := uint32(54000 + os.Getpid()%1000)
 	dsn = fmt.Sprintf("postgres://postgres:postgres@localhost:%d/postgres?sslmode=disable", port)
 
+	rt := filepath.Join(os.TempDir(), fmt.Sprintf("tapir-epg-store-%d", os.Getpid()))
 	pg := embeddedpostgres.NewDatabase(
-		embeddedpostgres.DefaultConfig().Port(port),
+		embeddedpostgres.DefaultConfig().
+			Port(port).
+			RuntimePath(rt).
+			DataPath(filepath.Join(rt, "data")).
+			BinariesPath(filepath.Join(rt, "bin")).
+			CachePath(filepath.Join(os.TempDir(), "tapir-epg-cache")),
 	)
 	if err := pg.Start(); err != nil {
 		fmt.Fprintf(os.Stderr, "embedded-postgres start: %v\n", err)
@@ -40,6 +52,7 @@ func TestMain(m *testing.M) {
 	if err := pg.Stop(); err != nil {
 		fmt.Fprintf(os.Stderr, "embedded-postgres stop: %v\n", err)
 	}
+	_ = os.RemoveAll(rt)
 	os.Exit(code)
 }
 

internal/adapters/store/videos.go

@@ -46,15 +46,16 @@ func (s *Store) UpsertVideo(ctx context.Context, v domain.Video) (string, error)
 		}
 
 		if err := tx.QueryRow(ctx,
-			`INSERT INTO videos (user_id, provider, provider_video_id, title, url, published_at, channel_title)
+			`INSERT INTO videos (user_id, provider, provider_video_id, title, url, published_at, channel_title, duration_s)
-			 VALUES ($1, $2, $3, $4, $5, $6, $7)
+			 VALUES ($1, $2, $3, $4, $5, $6, $7, $8)
 			 ON CONFLICT (user_id, provider, provider_video_id) DO UPDATE SET
 			     title         = EXCLUDED.title,
 			     url           = EXCLUDED.url,
 			     published_at  = EXCLUDED.published_at,
-			     channel_title = COALESCE(NULLIF(EXCLUDED.channel_title, ''), videos.channel_title)
+			     channel_title = COALESCE(NULLIF(EXCLUDED.channel_title, ''), videos.channel_title),
+			     duration_s    = COALESCE(EXCLUDED.duration_s, videos.duration_s)
 			 RETURNING id`,
-			v.UserID, provider, v.ProviderVideoID, v.Title, v.URL, nullTime(v.PublishedAt), v.ChannelTitle,
+			v.UserID, provider, v.ProviderVideoID, v.Title, v.URL, nullTime(v.PublishedAt), v.ChannelTitle, nullDuration(v.DurationSeconds),
 		).Scan(&id); err != nil {
 			return fmt.Errorf("store: upsert video: %w", err)
 		}
@@ -74,12 +75,27 @@ func nullTime(t time.Time) *time.Time {
 	return &t
 }
 
-// NewestUnsummarizedVideoIDs returns up to limit of the user's videos that have
+// nullDuration maps an unknown duration (0) to SQL NULL so the upsert's
-// no summary yet, newest first (published_at DESC, NULLS LAST). It caps the
+// COALESCE(EXCLUDED.duration_s, videos.duration_s) preserves a previously-known
-// connect-time onboarding burst (Feature 1) at a fixed count: the caller marks
+// value instead of clobbering it with 0 (ADR-028; the channel_title backfill
-// these for summarization through the shared rate gate. RLS-scoped via withUser,
+// stance, migration 014).
-// so it only ever sees the requesting user's rows. limit <= 0 returns nil.
+func nullDuration(seconds int) *int {
-func (s *Store) NewestUnsummarizedVideoIDs(ctx context.Context, userID string, limit int) ([]string, error) {
+	if seconds <= 0 {
+		return nil
+	}
+	return &seconds
+}
+
+// OnboardBurstVideoIDs returns up to limit of the user's unsummarized videos for
+// the connect-time onboarding burst (ADR-028), newest-first but quality-aware: a
+// video is excluded when its duration is KNOWN and outside [minSeconds, maxSeconds]
+// — dropping Shorts (below min) and multi-hour livestream VODs (above max) that
+// would waste a scarce caption fetch on a poor first impression. A NULL/unknown
+// duration is kept (degrade-open) but ranked AFTER known-good rows, so a freshly
+// enriched good pick wins when both exist. minSeconds<=0 / maxSeconds<=0 each
+// disable that bound (0/0 == pure newest-first, the reversibility lever).
+// RLS-scoped via withUser; limit <= 0 returns nil.
+func (s *Store) OnboardBurstVideoIDs(ctx context.Context, userID string, limit, minSeconds, maxSeconds int) ([]string, error) {
 	if limit <= 0 {
 		return nil, nil
 	}
@@ -92,16 +108,21 @@ func (s *Store) NewestUnsummarizedVideoIDs(ctx context.Context, userID string, l
 			    AND NOT EXISTS (
 			          SELECT 1 FROM summaries su
 			           WHERE su.user_id = v.user_id AND su.video_id = v.id)
-			  ORDER BY v.published_at DESC NULLS LAST, v.seen_at DESC
+			    AND NOT (
-			  LIMIT $2`, userID, limit)
+			          v.duration_s IS NOT NULL
+			          AND ( ($3 > 0 AND v.duration_s < $3)
+			             OR ($4 > 0 AND v.duration_s > $4) ))
+			  ORDER BY (v.duration_s IS NOT NULL) DESC,
+			           v.published_at DESC NULLS LAST, v.seen_at DESC
+			  LIMIT $2`, userID, limit, minSeconds, maxSeconds)
 		if err != nil {
-			return fmt.Errorf("store: newest unsummarized: %w", err)
+			return fmt.Errorf("store: onboard burst videos: %w", err)
 		}
 		defer rows.Close()
 		for rows.Next() {
 			var id string
 			if err := rows.Scan(&id); err != nil {
-				return fmt.Errorf("store: scan newest unsummarized: %w", err)
+				return fmt.Errorf("store: scan onboard burst video: %w", err)
 			}
 			ids = append(ids, id)
 		}

internal/adapters/store/videos_test.go

@@ -52,6 +52,36 @@ func TestUpsertVideo_ReturnsStableID(t *testing.T) {
 	require.Equal(t, 1, count, "must not duplicate the row")
 }
 
+func TestUpsertVideo_PersistsAndPreservesDuration(t *testing.T) {
+	ctx := context.Background()
+	s := newStore(t)
+	resetDB(t, rawPool(t))
+
+	// First upsert carries a known duration (ADR-028: discovery enriches it).
+	v := ytVideo(userA, "dur0000001x", "with duration")
+	v.DurationSeconds = 750
+	id, err := s.UpsertVideo(ctx, v)
+	require.NoError(t, err)
+
+	p := rawPool(t)
+	readDuration := func() *int {
+		var d *int
+		require.NoError(t, p.QueryRow(ctx, `SELECT duration_s FROM videos WHERE id = $1`, id).Scan(&d))
+		return d
+	}
+	require.NotNil(t, readDuration())
+	require.Equal(t, 750, *readDuration(), "duration must persist")
+
+	// A later upsert that does NOT know the duration (0) must not clobber it —
+	// the channel_title backfill stance (migration 014): COALESCE-preserve.
+	v2 := ytVideo(userA, "dur0000001x", "title updated, duration unknown")
+	v2.DurationSeconds = 0
+	_, err = s.UpsertVideo(ctx, v2)
+	require.NoError(t, err)
+	require.NotNil(t, readDuration(), "a 0/unknown re-upsert must not erase a known duration")
+	require.Equal(t, 750, *readDuration())
+}
+
 func TestUpsertVideo_IDMatchesSummaryDedup(t *testing.T) {
 	ctx := context.Background()
 	s := newStore(t)
@@ -82,36 +112,55 @@ func TestUpsertVideo_PerUserIsolation(t *testing.T) {
 	require.NotEqual(t, idA, idB, "same provider video for two users must be two distinct rows")
 }
 
-func TestNewestUnsummarizedVideoIDs(t *testing.T) {
+func TestOnboardBurstVideoIDs(t *testing.T) {
 	ctx := context.Background()
 	s := newStore(t)
 	resetDB(t, rawPool(t))
 
-	mk := func(user, pid string, day int) string {
+	mk := func(user, pid string, day, dur int) string {
 		v := ytVideo(user, pid, pid)
 		v.PublishedAt = time.Date(2026, 6, day, 12, 0, 0, 0, time.UTC)
+		v.DurationSeconds = dur // 0 == unknown (NULL)
 		id, err := s.UpsertVideo(ctx, v)
 		require.NoError(t, err)
 		return id
 	}
 
-	_ = mk(userA, "a1vid000001", 1)
+	summarized := mk(userA, "summ0000001", 6, 600) // newest known-good, but already summarized
-	id2 := mk(userA, "a2vid000002", 2)
+	good1 := mk(userA, "good0000001", 5, 600)      // 10m, newest UNsummarized known-good
-	id3 := mk(userA, "a3vid000003", 3)
+	tooLong := mk(userA, "toolong0001", 4, 20000)  // > maxSeconds -> dropped
-	id4 := mk(userA, "a4vid000004", 4)
+	tooShort := mk(userA, "tooshort001", 3, 30)    // < minSeconds -> dropped
-	mk(userB, "b1vid000009", 9) // userB's newest — must never leak via RLS
+	unknown := mk(userA, "unknown0001", 2, 0)      // NULL duration -> kept, ranked last
+	good2 := mk(userA, "good0000002", 1, 800)      // known-good but oldest
+	mk(userB, "bvid0000009", 9, 600)               // userB -> must not leak via RLS
 
-	// The newest (v4) is summarized, so it's excluded from "unsummarized".
+	// The newest video is summarized, so it is excluded from the burst.
-	require.NoError(t, s.Deliver(ctx, summary(userA, id4, "done")))
+	require.NoError(t, s.Deliver(ctx, summary(userA, summarized, "done")))
 
-	// Cap 2, newest-first unsummarized: v3 then v2 (v4 excluded; userB excluded).
+	const minSec, maxSec = 60, 14400
-	got, err := s.NewestUnsummarizedVideoIDs(ctx, userA, 2)
+
+	// Known-good ranked before unknown, each newest-first within its group; the
+	// too-long and too-short videos are excluded by their known duration.
+	got, err := s.OnboardBurstVideoIDs(ctx, userA, 5, minSec, maxSec)
 	require.NoError(t, err)
-	require.Equal(t, []string{id3, id2}, got)
+	require.Equal(t, []string{good1, good2, unknown}, got,
+		"known-good first (newest-first), then unknown-duration; junk excluded")
 
-	none, err := s.NewestUnsummarizedVideoIDs(ctx, userA, 0)
+	// Cap is honoured.
+	capped, err := s.OnboardBurstVideoIDs(ctx, userA, 2, minSec, maxSec)
 	require.NoError(t, err)
-	require.Empty(t, none, "limit 0 returns nothing")
+	require.Equal(t, []string{good1, good2}, capped)
+
+	// Bounds disabled (0/0) == pure newest-first, nothing excluded.
+	all, err := s.OnboardBurstVideoIDs(ctx, userA, 10, 0, 0)
+	require.NoError(t, err)
+	require.ElementsMatch(t, []string{good1, tooLong, tooShort, unknown, good2}, all,
+		"0/0 bounds disable the duration filter (prior newest-first behaviour)")
+
+	// limit <= 0 returns nothing.
+	none, err := s.OnboardBurstVideoIDs(ctx, userA, 0, minSec, maxSec)
+	require.NoError(t, err)
+	require.Empty(t, none)
 }
 
 func TestUpsertVideoPersistsChannelAndDistinctChannels(t *testing.T) {

internal/adapters/youtube/gate.go

@@ -2,6 +2,7 @@ package youtube
 
 import (
 	"context"
+	"sync/atomic"
 	"time"
 
 	"golang.org/x/time/rate"
@@ -29,9 +30,55 @@ func SetFetchRate(interval time.Duration) {
 	globalFetchGate = rate.NewLimiter(rate.Every(interval), 1)
 }
 
+// foregroundPending counts in-flight foreground (user-initiated) caption fetches.
+// The background sweep yields the gate while this is non-zero so a human waiting
+// on a click gets the next slot — and, on a near-throttled IP, the pre-429 window
+// — instead of competing equally with the firehose (ADR-026, Pillar A). Clicks are
+// rare and bursty, so background barely notices; the win to the click is large.
+var foregroundPending atomic.Int64
+
+// fgCtxKey marks a context as foreground (user-initiated). Unexported; set via
+// ForegroundContext and read via isForeground so only this package owns the key.
+type fgCtxKey struct{}
+
+// ForegroundContext marks ctx as a user-initiated (foreground) fetch so the gate
+// gives it priority. The web "Summarize"/paste/retry path wraps its context with
+// this; the background scheduler leaves it unset.
+func ForegroundContext(ctx context.Context) context.Context {
+	return context.WithValue(ctx, fgCtxKey{}, true)
+}
+
+func isForeground(ctx context.Context) bool {
+	v, _ := ctx.Value(fgCtxKey{}).(bool)
+	return v
+}
+
+// fgYieldPoll is how often a background waiter re-checks whether a foreground
+// fetch is still pending. Short enough to feel immediate, long enough not to spin.
+const fgYieldPoll = 200 * time.Millisecond
+
 // 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.
+//
+// Foreground (user-initiated) fetches take priority: they register as pending and
+// acquire a token immediately. Background fetches first yield — they wait until no
+// foreground fetch is pending — so a live click is never stuck behind the
+// background sweep and gets the cleaner slot against the per-IP limit (ADR-026).
 func WaitFetchGate(ctx context.Context) error {
+	if isForeground(ctx) {
+		foregroundPending.Add(1)
+		defer foregroundPending.Add(-1)
+		return globalFetchGate.Wait(ctx)
+	}
+
+	// Background: defer to any pending foreground fetch before taking a token.
+	for foregroundPending.Load() > 0 {
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		case <-time.After(fgYieldPoll):
+		}
+	}
 	return globalFetchGate.Wait(ctx)
 }

internal/adapters/youtube/gate_test.go

@@ -82,3 +82,37 @@ func TestSetFetchRateZeroIsUnlimited(t *testing.T) {
 		require.NoError(t, WaitFetchGate(context.Background()))
 	}
 }
+
+func TestForegroundContextMarker(t *testing.T) {
+	require.False(t, isForeground(context.Background()), "plain context is background")
+	require.True(t, isForeground(ForegroundContext(context.Background())), "marked context is foreground")
+}
+
+// TestWaitFetchGateForegroundProceedsImmediately: a foreground fetch acquires a
+// token without yielding, even when background callers exist.
+func TestWaitFetchGateForegroundProceedsImmediately(t *testing.T) {
+	SetFetchRate(0) // unlimited limiter — isolate the yield logic from pacing
+	foregroundPending.Store(0)
+	t.Cleanup(func() { foregroundPending.Store(0) })
+
+	ctx, cancel := context.WithTimeout(context.Background(), time.Second)
+	defer cancel()
+	require.NoError(t, WaitFetchGate(ForegroundContext(ctx)), "foreground proceeds immediately")
+}
+
+// TestWaitFetchGateBackgroundYieldsToForeground: while a foreground fetch is
+// pending, a background fetch yields (does not take a token) until the foreground
+// clears — proven by a background wait timing out against its own deadline, then
+// succeeding once the foreground is done.
+func TestWaitFetchGateBackgroundYieldsToForeground(t *testing.T) {
+	SetFetchRate(0)
+	foregroundPending.Store(1) // simulate a foreground fetch in flight
+	t.Cleanup(func() { foregroundPending.Store(0) })
+
+	ctx, cancel := context.WithTimeout(context.Background(), 250*time.Millisecond)
+	defer cancel()
+	require.Error(t, WaitFetchGate(ctx), "background yields (blocks) while foreground is pending")
+
+	foregroundPending.Store(0) // foreground done
+	require.NoError(t, WaitFetchGate(context.Background()), "background proceeds once foreground clears")
+}

internal/adapters/youtube/youtube.go

@@ -302,6 +302,10 @@ func (a *Adapter) filterLowValue(ctx context.Context, client *http.Client, video
 		if m.seconds > 0 && m.seconds < a.cfg.MinVideoSeconds {
 			continue // Short / sub-threshold clip
 		}
+		// Carry the duration we already fetched onto the kept video so the store
+		// can persist it (ADR-028) — the burst's length-aware selection depends on
+		// it. Discarding it here was the gap the onboarding investigation found.
+		v.DurationSeconds = m.seconds
 		kept = append(kept, v)
 	}
 	return kept

internal/adapters/youtube/youtube_test.go

@@ -224,6 +224,11 @@ func TestNewVideosFiltersShortsAndLive(t *testing.T) {
 	if len(vids) != 1 || vids[0].ProviderVideoID != "long1" {
 		t.Fatalf("expected only long1 to survive the filter, got %+v", vids)
 	}
+	// The duration fetched for the filter is carried onto the kept video so the
+	// store can persist it (ADR-028) instead of discarding it.
+	if vids[0].DurationSeconds != 750 {
+		t.Fatalf("kept video DurationSeconds = %d, want 750 (PT12M30S)", vids[0].DurationSeconds)
+	}
 }
 
 // TestNewVideosNoFilterWhenDisabled: MinVideoSeconds=0 keeps the pre-ADR-023

internal/config/config.go

@@ -33,7 +33,10 @@ type Config struct {
 	SummarizerModel string
 	// FallbackModel is the LOCAL fallback alias tried when the primary fails or
 	// returns unparseable output (ADR-022). Kept local so content stays on the
-	// homelab stack. Empty disables it. Default a bigger-context local model.
+	// homelab stack. Default is an IGUANA model (not koala) so the fallback runs
+	// on a different host than the koala primary — koala carries other loads, and
+	// a different host also means a different egress IP for the (rare) fallback.
+	// Empty disables it.
 	FallbackModel string
 	// CloudFallbackModel is the worst-case EXTERNAL fallback alias, tried only
 	// after every local endpoint has failed (ADR-022). For client deployments set
@@ -117,6 +120,21 @@ type Config struct {
 	// caption rate gate (ADR-014) — the cap bounds count, never the pacing. Default 3.
 	OnboardSummarizeCount int
 
+	// OnboardSummarizerModel is the summarizer alias the connect-time burst leads
+	// its chain with (ADR-028) — a stronger model is affordable on the ≤3 summaries
+	// that form a new user's first impression. It heads a burst-specific chain;
+	// the standard chain (ADR-022) follows as resilience. Empty (or equal to
+	// SummarizerModel) collapses the burst back onto the shared processor — the
+	// reversibility lever. Default iguana/gemma4-26b (the brain-validated model).
+	OnboardSummarizerModel string
+
+	// OnboardMaxVideoSeconds upper-bounds the duration of a video the onboarding
+	// burst will pick (ADR-028), so the burst does not spend a scarce caption fetch
+	// on a multi-hour livestream VOD that passed the live filter once it ended. Only
+	// a KNOWN duration outside [MinVideoSeconds, this] is dropped; a NULL/unknown
+	// duration is kept (degrade-open). 0 disables the upper bound. Default 14400 (4h).
+	OnboardMaxVideoSeconds int
+
 	// 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.
@@ -146,26 +164,28 @@ func (c Config) DexConfigured() bool { return strings.TrimSpace(c.OIDCIssuer) !=
 
 // Defaults (see docs/homelab-integration.md). All overridable via env.
 const (
-	defaultGatewayURL            = "http://koala:30401/v1"
+	defaultGatewayURL             = "http://koala:30401/v1"
-	defaultSummarizerModel       = "koala/phi4-mini"
+	defaultSummarizerModel        = "koala/phi4-mini"
-	defaultFallbackModel         = "koala/phi4-14b"
+	defaultFallbackModel          = "iguana/gemma4-26b"
-	defaultCloudFallbackModel    = "berget/mistral-small"
+	defaultCloudFallbackModel     = "berget/mistral-small"
-	defaultSummaryMaxTokens      = 1500
+	defaultSummaryMaxTokens       = 1500
-	defaultMaxTranscriptChars    = 18000
+	defaultMaxTranscriptChars     = 18000
-	defaultMinVideoSeconds       = 60
+	defaultMinVideoSeconds        = 60
-	defaultCaptionlessThreshold  = 5
+	defaultCaptionlessThreshold   = 5
-	defaultCaptionlessWindow     = 14 * 24 * time.Hour
+	defaultCaptionlessWindow      = 14 * 24 * time.Hour
-	defaultSummarizerTimeout     = 5 * time.Minute
+	defaultSummarizerTimeout      = 5 * time.Minute
-	defaultYTTokenRef            = "youtube/refresh_token"
+	defaultYTTokenRef             = "youtube/refresh_token"
-	defaultYTConnectRedirectURL  = "https://tapir.d-ma.be/oauth/youtube/callback"
+	defaultYTConnectRedirectURL   = "https://tapir.d-ma.be/oauth/youtube/callback"
-	defaultOAuthRedirectAddr     = "localhost:8080"
+	defaultOAuthRedirectAddr      = "localhost:8080"
-	defaultHTTPAddr              = ":8080"
+	defaultHTTPAddr               = ":8080"
-	defaultFetchBackoff          = time.Hour
+	defaultFetchBackoff           = time.Hour
-	defaultFetchRate             = 2 * time.Second
+	defaultFetchRate              = 2 * time.Second
-	defaultPublicURL             = "https://tapir.d-ma.be"
+	defaultPublicURL              = "https://tapir.d-ma.be"
-	defaultAutoSummarizeWindow   = 7 * 24 * time.Hour
+	defaultAutoSummarizeWindow    = 7 * 24 * time.Hour
-	defaultOnboardSummarizeCount = 3
+	defaultOnboardSummarizeCount  = 3
-	maxOnboardSummarizeCount     = 5
+	maxOnboardSummarizeCount      = 5
+	defaultOnboardSummarizerModel = "iguana/gemma4-26b"
+	defaultOnboardMaxVideoSeconds = 14400 // 4h
 )
 
 // Load reads the environment into a Config, applying defaults. It does not
@@ -174,26 +194,27 @@ const (
 // it needs.
 func Load() (Config, error) {
 	c := Config{
-		UserID:               os.Getenv("TAPIR_USER_ID"),
+		UserID:                 os.Getenv("TAPIR_USER_ID"),
-		GatewayURL:           envOr("TAPIR_GATEWAY_URL", defaultGatewayURL),
+		GatewayURL:             envOr("TAPIR_GATEWAY_URL", defaultGatewayURL),
-		GatewayKey:           os.Getenv("TAPIR_GATEWAY_KEY"),
+		GatewayKey:             os.Getenv("TAPIR_GATEWAY_KEY"),
-		SummarizerModel:      envOr("TAPIR_SUMMARIZER_MODEL", defaultSummarizerModel),
+		SummarizerModel:        envOr("TAPIR_SUMMARIZER_MODEL", defaultSummarizerModel),
-		FallbackModel:        lookupOr("TAPIR_FALLBACK_MODEL", defaultFallbackModel),
+		OnboardSummarizerModel: lookupOr("TAPIR_ONBOARD_SUMMARIZER_MODEL", defaultOnboardSummarizerModel),
-		CloudFallbackModel:   lookupOr("TAPIR_CLOUD_FALLBACK_MODEL", defaultCloudFallbackModel),
+		FallbackModel:          lookupOr("TAPIR_FALLBACK_MODEL", defaultFallbackModel),
-		DBDSN:                os.Getenv("TAPIR_DB_DSN"),
+		CloudFallbackModel:     lookupOr("TAPIR_CLOUD_FALLBACK_MODEL", defaultCloudFallbackModel),
-		YTClientID:           os.Getenv("TAPIR_YT_CLIENT_ID"),
+		DBDSN:                  os.Getenv("TAPIR_DB_DSN"),
-		YTClientSecret:       os.Getenv("TAPIR_YT_CLIENT_SECRET"),
+		YTClientID:             os.Getenv("TAPIR_YT_CLIENT_ID"),
-		YTTokenRef:           envOr("TAPIR_YT_TOKEN_REF", defaultYTTokenRef),
+		YTClientSecret:         os.Getenv("TAPIR_YT_CLIENT_SECRET"),
-		YTConnectRedirectURL: envOr("TAPIR_YT_CONNECT_REDIRECT_URL", defaultYTConnectRedirectURL),
+		YTTokenRef:             envOr("TAPIR_YT_TOKEN_REF", defaultYTTokenRef),
-		SecretsFile:          envOr("TAPIR_SECRETS_FILE", defaultSecretsFile()),
+		YTConnectRedirectURL:   envOr("TAPIR_YT_CONNECT_REDIRECT_URL", defaultYTConnectRedirectURL),
-		OAuthRedirectAddr:    envOr("TAPIR_OAUTH_REDIRECT_ADDR", defaultOAuthRedirectAddr),
+		SecretsFile:            envOr("TAPIR_SECRETS_FILE", defaultSecretsFile()),
-		HTTPAddr:             envOr("TAPIR_HTTP_ADDR", defaultHTTPAddr),
+		OAuthRedirectAddr:      envOr("TAPIR_OAUTH_REDIRECT_ADDR", defaultOAuthRedirectAddr),
-		PublicURL:            envOr("TAPIR_PUBLIC_URL", defaultPublicURL),
+		HTTPAddr:               envOr("TAPIR_HTTP_ADDR", defaultHTTPAddr),
-		OIDCIssuer:           os.Getenv("TAPIR_OIDC_ISSUER"),
+		PublicURL:              envOr("TAPIR_PUBLIC_URL", defaultPublicURL),
-		DexClientID:          os.Getenv("TAPIR_DEX_CLIENT_ID"),
+		OIDCIssuer:             os.Getenv("TAPIR_OIDC_ISSUER"),
-		DexClientSecret:      os.Getenv("TAPIR_DEX_CLIENT_SECRET"),
+		DexClientID:            os.Getenv("TAPIR_DEX_CLIENT_ID"),
-		OIDCRedirectURL:      os.Getenv("TAPIR_OIDC_REDIRECT_URL"),
+		DexClientSecret:        os.Getenv("TAPIR_DEX_CLIENT_SECRET"),
-		SessionSecret:        os.Getenv("TAPIR_SESSION_SECRET"),
+		OIDCRedirectURL:        os.Getenv("TAPIR_OIDC_REDIRECT_URL"),
+		SessionSecret:          os.Getenv("TAPIR_SESSION_SECRET"),
 	}
 
 	timeout, err := durationOr("TAPIR_SUMMARIZER_TIMEOUT", defaultSummarizerTimeout)
@@ -283,6 +304,15 @@ func Load() (Config, error) {
 	}
 	c.OnboardSummarizeCount = onboard
 
+	onboardMax, err := intOr("TAPIR_ONBOARD_MAX_VIDEO_SECONDS", defaultOnboardMaxVideoSeconds)
+	if err != nil {
+		return Config{}, err
+	}
+	if onboardMax < 0 {
+		onboardMax = 0
+	}
+	c.OnboardMaxVideoSeconds = onboardMax
+
 	return c, nil
 }
 

internal/config/config_test.go

@@ -214,3 +214,62 @@ func TestValidateForAuth_PassesWhenComplete(t *testing.T) {
 		t.Errorf("ValidateForAuth: unexpected error %v", err)
 	}
 }
+
+func TestLoad_OnboardSummarizerModel(t *testing.T) {
+	cases := []struct {
+		name, env string
+		set       bool
+		want      string
+	}{
+		{"default", "", false, defaultOnboardSummarizerModel},
+		{"explicit", "koala/some-model", true, "koala/some-model"},
+		{"empty disables (collapses to shared processor)", "", true, ""},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			env := map[string]string{}
+			if c.set {
+				env["TAPIR_ONBOARD_SUMMARIZER_MODEL"] = c.env
+			}
+			setEnv(t, env)
+			cfg, err := Load()
+			if err != nil {
+				t.Fatalf("Load: %v", err)
+			}
+			if cfg.OnboardSummarizerModel != c.want {
+				t.Fatalf("OnboardSummarizerModel = %q, want %q", cfg.OnboardSummarizerModel, c.want)
+			}
+		})
+	}
+}
+
+func TestLoad_OnboardMaxVideoSeconds(t *testing.T) {
+	cases := []struct {
+		name, env string
+		want      int
+	}{
+		{"default", "", defaultOnboardMaxVideoSeconds},
+		{"explicit", "7200", 7200},
+		{"zero disables", "0", 0},
+		{"negative clamps to zero", "-9", 0},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			setEnv(t, map[string]string{"TAPIR_ONBOARD_MAX_VIDEO_SECONDS": c.env})
+			cfg, err := Load()
+			if err != nil {
+				t.Fatalf("Load: %v", err)
+			}
+			if cfg.OnboardMaxVideoSeconds != c.want {
+				t.Fatalf("OnboardMaxVideoSeconds = %d, want %d", cfg.OnboardMaxVideoSeconds, c.want)
+			}
+		})
+	}
+}
+
+func TestLoad_OnboardMaxVideoSecondsInvalid(t *testing.T) {
+	setEnv(t, map[string]string{"TAPIR_ONBOARD_MAX_VIDEO_SECONDS": "long"})
+	if _, err := Load(); err == nil {
+		t.Fatal("Load: want error for non-numeric TAPIR_ONBOARD_MAX_VIDEO_SECONDS")
+	}
+}

internal/domain/domain.go

@@ -77,6 +77,11 @@ type Video struct {
 	URL             string
 	PublishedAt     time.Time
 	SeenAt          time.Time
+	// DurationSeconds is the video length in seconds, when known (fetched by the
+	// ADR-023 videos.list enrichment at discovery). 0 means unknown — the store
+	// preserves a previously-known value rather than overwriting it with 0, and
+	// the onboarding burst (ADR-028) treats unknown as degrade-open (kept).
+	DurationSeconds int
 }
 
 // Transcript is the text of a video (or a record that none was available).

internal/web/chat.go

@@ -0,0 +1,227 @@
+package web
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"strings"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/chat"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+)
+
+// Chatter is the per-video chat backend (ADR-027). *chat.Service satisfies it;
+// tests substitute a fake. It carries NO caption-fetch dependency — the chat
+// handlers reach it only after reading an already-stored transcript, so an
+// enabled chat cannot trigger a fetch, touch the rate gate, or reach YouTube.
+type Chatter interface {
+	// Models returns the offerable models, local-first (cloud absent when disabled).
+	Models() []string
+	// DefaultModel resolves the model a fresh chat opens with given the summary's
+	// model (the ADR-027 default), falling back to the first offered model.
+	DefaultModel(summaryModel string) string
+	// Answer runs one chat turn against the supplied stored transcript text.
+	Answer(ctx context.Context, req chat.Request) (chat.Reply, error)
+}
+
+// maxHistoryTurns bounds the ephemeral conversation carried per request, so a long
+// back-and-forth cannot grow the prompt without limit (the transcript already
+// dominates the budget). Older turns drop off the front.
+const maxHistoryTurns = 8
+
+// chatView is everything the chat templates render: the video identity for links
+// and titles, the model switcher state, the running (ephemeral) conversation, and
+// the honest flags — Available is false when no usable transcript is stored
+// (ADR-027: honest "not available", never a fetch), Truncated when the transcript
+// was bounded to fit the model, Error for a transient model failure.
+type chatView struct {
+	VideoID   string
+	Title     string
+	Available bool
+	Models    []string
+	Selected  string
+	History   []chat.Turn
+	Truncated bool
+	Error     string
+}
+
+// handleChat renders the chat page for a summarized video (GET). Entry is scoped
+// through GetSummaryByVideo, which is RLS/user-scoped: a video that is not the
+// requesting user's own resolves to ErrNotFound → 404, so chat is reachable only
+// from the user's own summary view (ADR-027 isolation). The transcript is read
+// from the SHARED store (ADR-021) — a pure DB read, never a caption fetch; an
+// absent/text-less transcript renders the honest "not available" state, no fetch.
+func (a *App) handleChat(w http.ResponseWriter, r *http.Request) {
+	userID, ok := a.currentUserID(w, r)
+	if !ok {
+		return
+	}
+	row, ok := a.loadOwnedSummary(w, r, userID)
+	if !ok {
+		return
+	}
+	_, hasText, ok := a.readTranscript(w, r, *row)
+	if !ok {
+		return
+	}
+	view := chatView{
+		VideoID:   row.VideoID,
+		Title:     displayTitle(*row),
+		Available: hasText,
+		Models:    a.Chat.Models(),
+		Selected:  a.Chat.DefaultModel(row.AIModel),
+	}
+	// HTMX (the in-place reveal from the summary) gets just the open chat section,
+	// swapped over the closed dock so the summary above it stays put. A no-JS
+	// navigation gets the full page: the whole summary plus the open chat.
+	if isHTMX(r) {
+		a.render(w, r, chatSection(view))
+		return
+	}
+	a.render(w, r, ChatPage(*row, view))
+}
+
+// handleChatMessage answers one question against the stored transcript (POST).
+// It reads the transcript from the store (no fetch), runs the chosen model over
+// it plus the prior turns, appends the answer, and returns the refreshed chat
+// panel (HTMX) or the whole page (no-JS). A model failure is surfaced inline,
+// not as a 500 — the conversation and the question are preserved for a retry.
+func (a *App) handleChatMessage(w http.ResponseWriter, r *http.Request) {
+	userID, ok := a.currentUserID(w, r)
+	if !ok {
+		return
+	}
+	row, ok := a.loadOwnedSummary(w, r, userID)
+	if !ok {
+		return
+	}
+	if err := r.ParseForm(); err != nil {
+		http.Error(w, "bad form", http.StatusBadRequest)
+		return
+	}
+	transcript, hasText, ok := a.readTranscript(w, r, *row)
+	if !ok {
+		return
+	}
+
+	model := a.resolveModel(r.FormValue("model"), row.AIModel)
+	history := parseHistory(r.Form["hq"], r.Form["ha"])
+	question := strings.TrimSpace(r.FormValue("question"))
+
+	view := chatView{
+		VideoID:   row.VideoID,
+		Title:     displayTitle(*row),
+		Available: hasText,
+		Models:    a.Chat.Models(),
+		Selected:  model,
+		History:   history,
+	}
+
+	switch {
+	case !hasText:
+		// Honest "not available" — no fetch, no model call (ADR-027).
+	case question == "":
+		// A model switch with no text just re-renders — no wasted round-trip.
+	default:
+		reply, err := a.Chat.Answer(r.Context(), chat.Request{
+			Model:      model,
+			Transcript: transcript,
+			History:    history,
+			Question:   question,
+		})
+		if err != nil {
+			a.logger().Error("chat answer", "video", row.VideoID, "model", model, "err", err)
+			view.Error = "That model couldn't answer just now. Try again, or switch models."
+		} else {
+			view.History = appendTurn(history, chat.Turn{Question: question, Answer: reply.Answer})
+			view.Truncated = reply.Truncated
+		}
+	}
+	a.renderChatTurn(w, r, *row, view)
+}
+
+// loadOwnedSummary fetches the summary for the path's video scoped to userID, or
+// writes the right response (404 on not-found/not-owned, 500 on error) and reports
+// false. It is the single isolation gate for both chat handlers.
+func (a *App) loadOwnedSummary(w http.ResponseWriter, r *http.Request, userID string) (*store.SummaryRow, bool) {
+	videoID := r.PathValue("videoId")
+	row, err := a.Store.GetSummaryByVideo(r.Context(), userID, videoID)
+	if errors.Is(err, store.ErrNotFound) {
+		http.NotFound(w, r)
+		return nil, false
+	}
+	if err != nil {
+		a.serverError(w, r, "chat get summary", err)
+		return nil, false
+	}
+	return row, true
+}
+
+// readTranscript reads the shared stored transcript for a row (ADR-021) and
+// reports whether it carries usable text. It is a pure DB read — NO caption fetch,
+// the property the whole feature's safety rests on. ok is false only on a store
+// error (after a 500 is written); a missing/text-less transcript is (",", false,
+// true) — the honest "not available" case, handled by the caller, not an error.
+func (a *App) readTranscript(w http.ResponseWriter, r *http.Request, row store.SummaryRow) (content string, hasText, ok bool) {
+	t, found, err := a.Store.GetTranscript(r.Context(), row.Channel, row.ProviderVideoID)
+	if err != nil {
+		a.serverError(w, r, "chat get transcript", err)
+		return "", false, false
+	}
+	if !found || !t.HasText() {
+		return "", false, true
+	}
+	return t.Content, true, true
+}
+
+// renderChatTurn returns the chat panel fragment for an HTMX answer (swapped in
+// place within the open dock), or the full chat page otherwise — the no-JS POST
+// re-renders the whole summary + open chat with the new turn.
+func (a *App) renderChatTurn(w http.ResponseWriter, r *http.Request, row store.SummaryRow, v chatView) {
+	if isHTMX(r) {
+		a.render(w, r, chatPanel(v))
+		return
+	}
+	a.render(w, r, ChatPage(row, v))
+}
+
+// resolveModel keeps the posted model only when it is an offered option; anything
+// else (a forged value, or a model dropped because cloud is disabled) falls back
+// to the default. The chat.Service enforces the same guard before the gateway;
+// this keeps the rendered switcher honest too.
+func (a *App) resolveModel(posted, summaryModel string) string {
+	for _, m := range a.Chat.Models() {
+		if m == posted {
+			return posted
+		}
+	}
+	return a.Chat.DefaultModel(summaryModel)
+}
+
+// parseHistory zips the parallel hidden hq/ha fields back into ordered turns,
+// keeping only the most recent maxHistoryTurns. net/url preserves the submission
+// order of repeated fields, so the pairing is stable.
+func parseHistory(qs, as []string) []chat.Turn {
+	n := len(qs)
+	if len(as) < n {
+		n = len(as)
+	}
+	turns := make([]chat.Turn, 0, n)
+	for i := 0; i < n; i++ {
+		turns = append(turns, chat.Turn{Question: qs[i], Answer: as[i]})
+	}
+	return capHistory(turns)
+}
+
+// appendTurn adds a completed exchange and re-bounds the conversation.
+func appendTurn(history []chat.Turn, t chat.Turn) []chat.Turn {
+	return capHistory(append(history, t))
+}
+
+// capHistory keeps the last maxHistoryTurns turns (drops the oldest).
+func capHistory(turns []chat.Turn) []chat.Turn {
+	if len(turns) <= maxHistoryTurns {
+		return turns
+	}
+	return turns[len(turns)-maxHistoryTurns:]
+}

internal/web/chat_handler_test.go

@@ -0,0 +1,375 @@
+package web_test
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"net/url"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/jackc/pgx/v5/pgxpool"
+	"github.com/stretchr/testify/require"
+
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/chat"
+	"gitea.d-ma.be/mathias/tapir/internal/adapters/store"
+	"gitea.d-ma.be/mathias/tapir/internal/domain"
+	"gitea.d-ma.be/mathias/tapir/internal/web"
+)
+
+// videoZ is a video id used by the isolation test for a DIFFERENT user's video.
+const videoZ = "33333333-3333-3333-3333-333333333333"
+
+// --- chat test doubles -----------------------------------------------------
+
+// fakeChatter is a web.Chatter that records the request it received and returns a
+// canned reply. It performs NO network and NO fetch — it stands in for the real
+// chat.Service so the handler behaviour is what's under test.
+type fakeChatter struct {
+	models    []string
+	reply     chat.Reply
+	err       error
+	gotReq    chat.Request
+	callCount int
+}
+
+func (f *fakeChatter) Models() []string { return f.models }
+
+func (f *fakeChatter) DefaultModel(summaryModel string) string {
+	for _, m := range f.models {
+		if m == summaryModel {
+			return summaryModel
+		}
+	}
+	if len(f.models) > 0 {
+		return f.models[0]
+	}
+	return ""
+}
+
+func (f *fakeChatter) Answer(_ context.Context, req chat.Request) (chat.Reply, error) {
+	f.callCount++
+	f.gotReq = req
+	if f.err != nil {
+		return chat.Reply{}, f.err
+	}
+	return f.reply, nil
+}
+
+// tripwireProcessor and tripwireFetcher are the YouTube-reaching collaborators
+// (summarize → caption fetch, and the paste metadata fetch). Wired into the App
+// for the safety test, they fail it the instant chat routes into either — the
+// behavioural proof that chat never triggers a fetch (ADR-027).
+type tripwireProcessor struct{ t *testing.T }
+
+func (p tripwireProcessor) ProcessVideo(context.Context, string, string) error {
+	p.t.Fatal("chat triggered summarization (→ caption fetch) — must never happen (ADR-027)")
+	return nil
+}
+
+type tripwireFetcher struct{ t *testing.T }
+
+func (f tripwireFetcher) FetchVideo(context.Context, string, string) (domain.Video, error) {
+	f.t.Fatal("chat triggered a YouTube video fetch — must never happen (ADR-027)")
+	return domain.Video{}, nil
+}
+
+// --- helpers ---------------------------------------------------------------
+
+// newChatApp builds the App under test as the registered stub user, with a Chat
+// backend wired. Mirrors newApp but adds chat (and any extra wiring via mutate).
+func newChatApp(t *testing.T, chatter web.Chatter, mutate func(*web.App)) *web.App {
+	t.Helper()
+	s := newStore(t)
+	app := &web.App{
+		Store:    s,
+		Identity: s,
+		Auth:     web.StubAuth{U: web.User{Subject: stubSubject}},
+		Chat:     chatter,
+	}
+	if mutate != nil {
+		mutate(app)
+	}
+	return app
+}
+
+// seededProviderVideoID mirrors seedVideo's derivation so the chat path's
+// (provider, providerVideoID) transcript key matches the seeded video row.
+func seededProviderVideoID(videoID string) string { return "pv-" + videoID[:8] }
+
+// seedTranscript stores a shared (provider, providerVideoID) transcript — the
+// ADR-021 stored content the chat reads. Captions source = usable text.
+func seedTranscript(t *testing.T, s *store.Store, videoID, content string) {
+	t.Helper()
+	err := s.SaveTranscript(context.Background(), "youtube", seededProviderVideoID(videoID),
+		domain.Transcript{Source: domain.SourceCaptions, Language: "en", Content: content})
+	require.NoError(t, err)
+}
+
+func getChat(t *testing.T, app *web.App, videoID string) *httptest.ResponseRecorder {
+	t.Helper()
+	return do(t, app, httptest.NewRequest(http.MethodGet, "/v/"+videoID+"/chat", nil))
+}
+
+func postChat(t *testing.T, app *web.App, videoID string, form url.Values, htmx bool) *httptest.ResponseRecorder {
+	t.Helper()
+	req := httptest.NewRequest(http.MethodPost, "/v/"+videoID+"/chat", strings.NewReader(form.Encode()))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	if htmx {
+		req.Header.Set("HX-Request", "true")
+	}
+	return do(t, app, req)
+}
+
+// --- scenarios -------------------------------------------------------------
+
+// The "dig deeper" affordance appears on a summary detail view only when chat is
+// enabled, and links to that video's chat — the single entry point (ADR-027 §1).
+func TestChatEntryAffordanceOnSummaryView(t *testing.T) {
+	ctx := context.Background()
+	withChat := newChatApp(t, &fakeChatter{models: []string{"phi4-mini"}}, nil)
+	p := rawPool(t)
+	resetDB(t, p)
+	require.NoError(t, deliver(ctx, withChat, videoX, "body x"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
+
+	html := body(t, do(t, withChat, httptest.NewRequest(http.MethodGet, "/v/"+videoX, nil)))
+	require.Contains(t, html, "Dig deeper", "the deeper-dive affordance is shown when chat is enabled")
+	require.Contains(t, html, "/v/"+videoX+"/chat", "it links to this video's chat")
+	require.Contains(t, html, `id="chat-section"`, "the dock lives on the detail page")
+	require.Contains(t, html, `hx-get="/v/`+videoX+`/chat"`, "it opens the chat in place (HTMX), not a navigation")
+
+	// With no chat backend wired the affordance is absent (routes unmounted).
+	noChat := newApp(t)
+	require.NoError(t, deliver(ctx, noChat, videoX, "body x"))
+	html = body(t, do(t, noChat, httptest.NewRequest(http.MethodGet, "/v/"+videoX, nil)))
+	require.NotContains(t, html, "Dig deeper", "no affordance when chat is disabled")
+}
+
+// The summary and the chat live together (the integrated UX): the no-JS chat page
+// renders the full summary alongside the chat, and the HTMX reveal returns just
+// the open chat section as a fragment so it docks in below the summary already on
+// screen — the summary is never navigated away from.
+func TestChatIntegratedWithSummaryOnSamePage(t *testing.T) {
+	ctx := context.Background()
+	chatter := &fakeChatter{models: []string{"phi4-mini", "gemma4-26b"}}
+	app := newChatApp(t, chatter, nil)
+	p := rawPool(t)
+	resetDB(t, p)
+	require.NoError(t, deliver(ctx, app, videoX, "SUMMARY-BODY-MARKER"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
+	seedTranscript(t, app.Store.(*store.Store), videoX, "the transcript")
+
+	// No-JS full page: the summary payload and the chat are on one page.
+	html := body(t, getChat(t, app, videoX))
+	require.Contains(t, html, "SUMMARY-BODY-MARKER", "the summary text is shown on the chat page")
+	require.Contains(t, html, "takeaway one", "takeaways shown alongside the chat")
+	require.Contains(t, html, "highlight one", "highlights shown alongside the chat")
+	require.Contains(t, html, "Ask about this video", "the chat sits on the same page as the summary")
+	require.Contains(t, html, `name="question"`, "the ask form is present")
+
+	// HTMX reveal: the open chat section ONLY (a fragment) — no full-page chrome and
+	// no duplicated summary, so it swaps in below the summary already rendered.
+	req := httptest.NewRequest(http.MethodGet, "/v/"+videoX+"/chat", nil)
+	req.Header.Set("HX-Request", "true")
+	frag := body(t, do(t, app, req))
+	require.NotContains(t, frag, "<html", "the reveal is a fragment, not a full page")
+	require.NotContains(t, frag, "SUMMARY-BODY-MARKER", "the reveal does not re-send the summary (it's already on screen)")
+	require.Contains(t, frag, `id="chat-section"`, "the fragment replaces the dock in place")
+	require.Contains(t, frag, `name="question"`, "the ask form is in the revealed section")
+}
+
+// THE KEY SAFETY ASSERTION (ADR-027): a chat answer is produced entirely from the
+// stored transcript — the model receives the stored text, and neither the
+// summarize→fetch path nor the YouTube fetch path is ever touched. The tripwire
+// collaborators t.Fatal the test if chat reaches them.
+func TestChatAnswersFromStoredTranscriptWithoutAnyFetch(t *testing.T) {
+	ctx := context.Background()
+	const transcript = "STORED-TRANSCRIPT-MARKER: the host explains attention budgets."
+	chatter := &fakeChatter{
+		models: []string{"phi4-mini"},
+		reply:  chat.Reply{Answer: "It is about attention budgets."},
+	}
+	app := newChatApp(t, chatter, func(a *web.App) {
+		a.Processor = tripwireProcessor{t} // fails the test if chat summarizes/fetches
+		a.Fetcher = tripwireFetcher{t}     // fails the test if chat fetches video metadata
+	})
+	p := rawPool(t)
+	resetDB(t, p)
+	require.NoError(t, deliver(ctx, app, videoX, "summary body"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
+	seedTranscript(t, app.Store.(*store.Store), videoX, transcript)
+
+	form := url.Values{"question": {"what is it about?"}, "model": {"phi4-mini"}}
+	rec := postChat(t, app, videoX, form, true)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Contains(t, body(t, rec), "It is about attention budgets.", "the answer is rendered")
+	require.Equal(t, 1, chatter.callCount, "the model was asked exactly once")
+	require.Equal(t, transcript, chatter.gotReq.Transcript,
+		"the model answered from the STORED transcript, not a fetched one")
+	// The tripwires never firing IS the no-fetch / no-YouTube proof.
+}
+
+// A video with no stored transcript yields an honest "not available" — and never
+// a fetch, never a model call (ADR-027 §2: do not add an on-demand-fetch path).
+func TestChatUnavailableWhenNoStoredTranscript(t *testing.T) {
+	ctx := context.Background()
+	chatter := &fakeChatter{models: []string{"phi4-mini"}}
+	app := newChatApp(t, chatter, func(a *web.App) {
+		a.Processor = tripwireProcessor{t}
+		a.Fetcher = tripwireFetcher{t}
+	})
+	p := rawPool(t)
+	resetDB(t, p)
+	require.NoError(t, deliver(ctx, app, videoX, "summary body"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
+	// NB: no seedTranscript — the transcript is absent.
+
+	rec := getChat(t, app, videoX)
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Contains(t, body(t, rec), "isn't available", "honest not-available copy")
+
+	// Asking anyway still triggers nothing: no model call, no fetch.
+	rec = postChat(t, app, videoX, url.Values{"question": {"hi"}, "model": {"phi4-mini"}}, true)
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Equal(t, 0, chatter.callCount, "no model call without a stored transcript")
+}
+
+// Chat is reachable ONLY from the user's own summary view: opening chat for a
+// video that belongs to another user 404s (the summary read is RLS-scoped), so a
+// guessed/arbitrary video id is not a chat surface (ADR-027 §5 isolation).
+func TestChatOnlyReachableForOwnVideo(t *testing.T) {
+	chatter := &fakeChatter{models: []string{"phi4-mini"}}
+	app := newChatApp(t, chatter, func(a *web.App) {
+		a.Processor = tripwireProcessor{t}
+		a.Fetcher = tripwireFetcher{t}
+	})
+	p := rawPool(t)
+	resetDB(t, p)
+
+	// A summarized video with a stored transcript owned by ANOTHER user.
+	const other = "22222222-2222-2222-2222-222222222222"
+	seedForeignSummaryWithTranscript(t, p, other, videoZ, "Foreign Title")
+
+	rec := getChat(t, app, videoZ)
+	require.Equal(t, http.StatusNotFound, rec.Code, "cannot open chat for another user's video")
+	require.Equal(t, 0, chatter.callCount, "no model call for a non-owned video")
+
+	rec = postChat(t, app, videoZ, url.Values{"question": {"hi"}, "model": {"phi4-mini"}}, true)
+	require.Equal(t, http.StatusNotFound, rec.Code, "cannot post chat to another user's video")
+}
+
+// The switcher offers exactly the backend's models, defaults to the summary's own
+// model, and a switch re-runs against the SAME transcript with the chosen model
+// (ADR-027 §3 — model-comparison instrumentation). A model dropped from the offer
+// set (e.g. cloud disabled) is not rendered.
+func TestChatModelSwitcherDefaultAndSwitch(t *testing.T) {
+	ctx := context.Background()
+	// The seeded summary's model is "phi4-mini" (see handlers_test summary()).
+	chatter := &fakeChatter{
+		models: []string{"phi4-mini", "gemma4-26b"}, // note: no cloud model offered
+		reply:  chat.Reply{Answer: "answer"},
+	}
+	app := newChatApp(t, chatter, nil)
+	p := rawPool(t)
+	resetDB(t, p)
+	require.NoError(t, deliver(ctx, app, videoX, "summary body"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
+	seedTranscript(t, app.Store.(*store.Store), videoX, "the transcript text")
+
+	html := body(t, getChat(t, app, videoX))
+	require.Contains(t, html, "gemma4-26b", "every offered model is in the switcher")
+	require.NotContains(t, html, "mistral", "a non-offered (cloud-disabled) model is absent")
+	require.Contains(t, html, `value="phi4-mini" selected`, "defaults to the summary's own model")
+
+	// Switching to gemma re-runs against the same stored transcript.
+	form := url.Values{"question": {"q"}, "model": {"gemma4-26b"}}
+	rec := postChat(t, app, videoX, form, true)
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Equal(t, "gemma4-26b", chatter.gotReq.Model, "the chosen model answers")
+	require.Equal(t, "the transcript text", chatter.gotReq.Transcript, "against the same transcript")
+}
+
+// A truncated transcript surfaces the honest bounded-context note (ADR-027 §2).
+func TestChatTruncationNoteShown(t *testing.T) {
+	ctx := context.Background()
+	chatter := &fakeChatter{
+		models: []string{"phi4-mini"},
+		reply:  chat.Reply{Answer: "answer", Truncated: true},
+	}
+	app := newChatApp(t, chatter, nil)
+	p := rawPool(t)
+	resetDB(t, p)
+	require.NoError(t, deliver(ctx, app, videoX, "summary body"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
+	seedTranscript(t, app.Store.(*store.Store), videoX, "long transcript")
+
+	rec := postChat(t, app, videoX, url.Values{"question": {"q"}, "model": {"phi4-mini"}}, true)
+	require.Contains(t, body(t, rec), "bounded portion", "the truncation note is shown")
+}
+
+// A multi-turn conversation is carried in the request (hidden fields), not the DB:
+// prior turns ride back into the next answer, and nothing is persisted (ADR-027 §4).
+func TestChatMultiTurnHistoryIsEphemeral(t *testing.T) {
+	ctx := context.Background()
+	chatter := &fakeChatter{models: []string{"phi4-mini"}, reply: chat.Reply{Answer: "second answer"}}
+	app := newChatApp(t, chatter, nil)
+	p := rawPool(t)
+	resetDB(t, p)
+	require.NoError(t, deliver(ctx, app, videoX, "summary body"))
+	seedVideo(t, p, videoX, "X Title", "https://x", time.Time{})
+	seedTranscript(t, app.Store.(*store.Store), videoX, "the transcript")
+
+	// Second turn posts the prior exchange as hidden history fields.
+	form := url.Values{
+		"question": {"follow-up question"},
+		"model":    {"phi4-mini"},
+		"hq":       {"first question"},
+		"ha":       {"first answer"},
+	}
+	rec := postChat(t, app, videoX, form, true)
+	require.Equal(t, http.StatusOK, rec.Code)
+
+	require.Len(t, chatter.gotReq.History, 1, "the prior turn was carried into the request")
+	require.Equal(t, "first question", chatter.gotReq.History[0].Question)
+	require.Equal(t, "first answer", chatter.gotReq.History[0].Answer)
+
+	html := body(t, rec)
+	require.Contains(t, html, "second answer", "the new answer renders")
+	require.Contains(t, html, "first question", "the conversation persists in page state")
+
+	// Nothing was written to the DB — there is no chat table; the summaries/actions
+	// are unchanged by a chat turn.
+	row, err := app.Store.GetSummaryByVideo(ctx, userID, videoX)
+	require.NoError(t, err)
+	require.Equal(t, "summary body", row.Summary, "chat never mutates stored data")
+}
+
+// --- foreign-user seeding (isolation) --------------------------------------
+
+// seedForeignSummaryWithTranscript creates a fully-summarized, transcript-backed
+// video owned by a DIFFERENT user via the raw pool (which bypasses RLS), so the
+// isolation test can confirm the requesting user cannot open chat for it.
+func seedForeignSummaryWithTranscript(t *testing.T, p *pgxpool.Pool, otherUserID, videoID, title string) {
+	t.Helper()
+	ctx := context.Background()
+	_, err := p.Exec(ctx, `INSERT INTO users (id) VALUES ($1) ON CONFLICT (id) DO NOTHING`, otherUserID)
+	require.NoError(t, err)
+	_, err = p.Exec(ctx,
+		`INSERT INTO videos (id, user_id, provider, provider_video_id, title, url)
+		 VALUES ($1, $2, 'youtube', $3, $4, 'https://z')`,
+		videoID, otherUserID, seededProviderVideoID(videoID), title)
+	require.NoError(t, err)
+	_, err = p.Exec(ctx,
+		`INSERT INTO summaries (user_id, video_id, summary, ai_provider, ai_model)
+		 VALUES ($1, $2, 'foreign summary', 'local', 'phi4-mini')`,
+		otherUserID, videoID)
+	require.NoError(t, err)
+	_, err = p.Exec(ctx,
+		`INSERT INTO transcripts (provider, provider_video_id, source, language, content)
+		 VALUES ('youtube', $1, 'captions', 'en', 'foreign transcript')`,
+		seededProviderVideoID(videoID))
+	require.NoError(t, err)
+}

internal/web/handlers.go

@@ -26,6 +26,10 @@ type Store interface {
 	DistinctChannels(ctx context.Context, userID string) ([]string, error)
 	GetSummaryByVideo(ctx context.Context, userID, videoID string) (*store.SummaryRow, error)
 	GetVideoRow(ctx context.Context, userID, videoID string) (*store.SummaryRow, error)
+	// GetTranscript reads the shared, stored transcript keyed by (provider,
+	// providerVideoID) — ADR-021. It is a pure DB read: it never fetches captions,
+	// so the chat path (ADR-027) reaches it without any caption-fetch surface.
+	GetTranscript(ctx context.Context, provider, providerVideoID string) (domain.Transcript, bool, 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
@@ -91,6 +95,11 @@ type App struct {
 	// Fetcher, when non-nil, resolves an arbitrary YouTube video id to metadata for
 	// the paste-a-URL flow (Feature 2). Nil = the /paste route is not mounted.
 	Fetcher VideoFetcher
+	// Chat, when non-nil, answers per-video questions against a video's STORED
+	// transcript (ADR-027). Nil = the /v/{id}/chat routes are not mounted and the
+	// summary view shows no "dig deeper" affordance. It holds no caption-fetch
+	// dependency, so an enabled chat cannot reach YouTube or the rate gate.
+	Chat Chatter
 	// 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
@@ -149,6 +158,12 @@ func (a *App) Router() http.Handler {
 		app.HandleFunc("POST /paste", a.handlePaste)
 	}
 	app.HandleFunc("GET /v/{videoId}/status", a.handleStatus)
+	// Per-video deeper-dive chat over the STORED transcript (ADR-027). Mounted only
+	// when a Chat backend is wired; it never fetches captions.
+	if a.Chat != nil {
+		app.HandleFunc("GET /v/{videoId}/chat", a.handleChat)
+		app.HandleFunc("POST /v/{videoId}/chat", a.handleChatMessage)
+	}
 	app.HandleFunc("GET /register", a.handleRegisterForm)
 	app.HandleFunc("POST /register", a.handleRegister)
 
@@ -265,7 +280,7 @@ func (a *App) handleDetail(w http.ResponseWriter, r *http.Request) {
 		a.serverError(w, r, "get summary", err)
 		return
 	}
-	a.render(w, r, DetailPage(*row))
+	a.render(w, r, DetailPage(*row, a.Chat != nil))
 }
 
 // handleAction toggles one action: re-clicking an active verb clears it, else it

internal/web/handlers_test.go

@@ -7,6 +7,7 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"os"
+	"path/filepath"
 	"strings"
 	"testing"
 	"time"
@@ -26,10 +27,22 @@ import (
 var dsn string
 
 func TestMain(m *testing.M) {
-	const port = 54330 // distinct from the store package's embedded PG (54329)
+	// Per-process port + dirs so concurrent `go test` runs (e.g. a push-run and a
+	// tag-run in CI) never collide on a fixed port or shared data dir. Base 55000
+	// keeps web's range distinct from the store package (54000). Shared CachePath
+	// downloads the PG archive once.
+	port := uint32(55000 + os.Getpid()%1000)
 	dsn = fmt.Sprintf("postgres://postgres:postgres@localhost:%d/postgres?sslmode=disable", port)
 
-	pg := embeddedpostgres.NewDatabase(embeddedpostgres.DefaultConfig().Port(port))
+	rt := filepath.Join(os.TempDir(), fmt.Sprintf("tapir-epg-web-%d", os.Getpid()))
+	pg := embeddedpostgres.NewDatabase(
+		embeddedpostgres.DefaultConfig().
+			Port(port).
+			RuntimePath(rt).
+			DataPath(filepath.Join(rt, "data")).
+			BinariesPath(filepath.Join(rt, "bin")).
+			CachePath(filepath.Join(os.TempDir(), "tapir-epg-cache")),
+	)
 	if err := pg.Start(); err != nil {
 		fmt.Fprintf(os.Stderr, "embedded-postgres start: %v\n", err)
 		os.Exit(1)
@@ -38,6 +51,7 @@ func TestMain(m *testing.M) {
 	if err := pg.Stop(); err != nil {
 		fmt.Fprintf(os.Stderr, "embedded-postgres stop: %v\n", err)
 	}
+	_ = os.RemoveAll(rt)
 	os.Exit(code)
 }
 

internal/web/view.go

@@ -184,6 +184,12 @@ func statusURL(videoID string) templ.SafeURL {
 	return templ.SafeURL("/v/" + videoID + "/status")
 }
 
+// chatURL builds the per-video chat path (GET renders the page, POST answers) —
+// the deeper-dive over the stored transcript (ADR-027).
+func chatURL(videoID string) templ.SafeURL {
+	return templ.SafeURL("/v/" + videoID + "/chat")
+}
+
 // 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.
@@ -725,6 +731,30 @@ a.btn, a.btn:visited { color: var(--accent-fg); }
 .detail ul { margin: 0; padding-left: 1.2rem; line-height: 1.6; }
 .detail li { margin-bottom: var(--s1); }
 
+/* deeper-dive chat (ADR-027) — docks in place below the summary */
+.chat-dock { margin-top: var(--s5); border-top: 1px solid var(--line); padding-top: var(--s4); }
+.chat-dock .chat-open { display: inline-block; }
+.chat-heading { font-size: 1.1rem; margin: 0 0 var(--s2); }
+.chat-scope { margin: 0 0 var(--s3); font-size: .9rem; }
+.chat-panel { display: flex; flex-direction: column; gap: var(--s3); }
+.chat-log { display: flex; flex-direction: column; gap: var(--s3); }
+.chat-turn { border-radius: var(--radius); padding: var(--s2) var(--s3); }
+.chat-turn p { margin: 0; }
+.chat-q { background: var(--accent-weak); color: var(--fg); align-self: flex-end; max-width: 85%; }
+.chat-a { background: var(--card); border: 1px solid var(--line); }
+.chat-a .body { white-space: pre-wrap; line-height: 1.6; }
+.chat-note { margin: 0; font-size: .82rem; font-style: italic; }
+.chat-error { margin: 0; color: #8a1c10; font-size: .9rem; }
+@media (prefers-color-scheme: dark) { .chat-error { color: #f3b5ae; } }
+.chat-form { display: flex; flex-direction: column; gap: var(--s2); margin: var(--s2) 0 0; }
+.chat-model { flex-direction: column; display: flex; gap: var(--s1); font-size: .78rem; text-transform: uppercase; letter-spacing: .04em; color: var(--muted); align-items: flex-start; }
+.chat-model select { font: inherit; text-transform: none; letter-spacing: 0; padding: .4rem .55rem; border: 1px solid var(--line); border-radius: var(--radius); background: var(--card); color: var(--fg); }
+.chat-model-hint { text-transform: none; letter-spacing: 0; font-size: .78rem; }
+.chat-form textarea { font: inherit; padding: .55rem; border: 1px solid var(--line); border-radius: var(--radius); background: var(--card); color: var(--fg); resize: vertical; }
+.chat-form textarea:focus-visible { outline: 2px solid var(--accent); outline-offset: 1px; border-color: var(--accent); }
+.chat-form .btn { align-self: flex-start; }
+.chat-thinking { font-style: italic; }
+
 /* action toggles — watched|skipped form one segmented control (they are mutually
    exclusive), "saved" sits apart as an independent toggle */
 .actions { display: flex; gap: var(--s3); margin: var(--s4) 0; flex-wrap: wrap; align-items: center; }

internal/web/views.templ

@@ -410,69 +410,175 @@ templ noCaptionsCard(r store.SummaryRow) {
 	</li>
 }
 
-// DetailPage is the full summary view: text, highlights, takeaways, metadata,
+// summaryBody is the summary payload shared by the detail page and the no-JS
-// and the action button group.
+// chat page (so the chat page shows the same summary, not a separate view):
-templ DetailPage(r store.SummaryRow) {
+// metadata, embed, source, the action toggles, then the attention-saving order
+// Takeaways → Highlights → Summary (UX review A8).
+templ summaryBody(r store.SummaryRow) {
+	<p class="meta">
+		if detailMeta(r) != "" {
+			<span>{ detailMeta(r) }</span>
+		}
+		if r.FallbackUsed {
+			<span class="badge" title="summarized with the fallback model" aria-label="summarized with the fallback model">fallback</span>
+		}
+	</p>
+	if url, ok := embedURL(r.ProviderVideoID); ok {
+		<div class="embed">
+			<iframe
+				src={ url }
+				title={ displayTitle(r) }
+				loading="lazy"
+				referrerpolicy="strict-origin-when-cross-origin"
+				allow="accelerometer; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
+				allowfullscreen
+			></iframe>
+		</div>
+	}
+	if r.URL != "" {
+		<p class="source"><a href={ externalURL(r.URL) } rel="noopener noreferrer">watch on source ↗</a></p>
+	}
+	@ActionButtons(r.VideoID, actionSet(r.Actions))
+	if len(r.Takeaways) > 0 {
+		<section>
+			<h2>Takeaways</h2>
+			<ul>
+				for _, t := range r.Takeaways {
+					<li>{ t }</li>
+				}
+			</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>
+}
+
+// DetailPage is the full summary view: the summary payload, then (when chat is
+// enabled) the deeper-dive dock (ADR-027) — a reveal that opens the chat IN PLACE
+// below the summary, so the summary stays on screen as the context being asked
+// about rather than being navigated away from.
+templ DetailPage(r store.SummaryRow, chatEnabled bool) {
 	@Layout("Tapir — " + displayTitle(r)) {
 		<article class="detail">
 			<p class="back"><a href="/">← Summaries</a></p>
 			<h1>{ displayTitle(r) }</h1>
-			<p class="meta">
+			@summaryBody(r)
-				if detailMeta(r) != "" {
+			if chatEnabled {
-					<span>{ detailMeta(r) }</span>
+				@chatReveal(r.VideoID)
-				}
-				if r.FallbackUsed {
-					<span class="badge" title="summarized with the fallback model" aria-label="summarized with the fallback model">fallback</span>
-				}
-			</p>
-			if url, ok := embedURL(r.ProviderVideoID); ok {
-				<div class="embed">
-					<iframe
-						src={ url }
-						title={ displayTitle(r) }
-						loading="lazy"
-						referrerpolicy="strict-origin-when-cross-origin"
-						allow="accelerometer; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
-						allowfullscreen
-					></iframe>
-				</div>
 			}
-			if r.URL != "" {
-				<p class="source"><a href={ externalURL(r.URL) } rel="noopener noreferrer">watch on source ↗</a></p>
-			}
-			@ActionButtons(r.VideoID, actionSet(r.Actions))
-			// Lead with the attention-saving payload: Takeaways ("is this worth my
-			// time?") first, then Highlights, then the full Summary last (UX review
-			// A8). Takeaways/Highlights are conditional, so a video without them falls
-			// through to the Summary leading naturally.
-			if len(r.Takeaways) > 0 {
-				<section>
-					<h2>Takeaways</h2>
-					<ul>
-						for _, t := range r.Takeaways {
-							<li>{ t }</li>
-						}
-					</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>
 	}
 }
 
+// chatReveal is the CLOSED dock at the foot of the summary: a quiet affordance,
+// not a loud CTA (it deepens value for a reader already here, never nudges). With
+// JS it swaps itself for the open chat section in place (HTMX, summary stays
+// above); without JS the same href navigates to the full chat page, which renders
+// the summary alongside the chat. Either way the summary is never lost.
+templ chatReveal(videoID string) {
+	<section id="chat-section" class="chat-dock">
+		<a
+			class="btn-secondary chat-open"
+			href={ chatURL(videoID) }
+			hx-get={ string(chatURL(videoID)) }
+			hx-target="#chat-section"
+			hx-swap="outerHTML"
+		>
+			Dig deeper — ask about this video →
+		</a>
+	</section>
+}
+
+// chatSection is the OPEN dock: heading + scope note + the chat panel, swapped in
+// over the closed reveal (same #chat-section id, outerHTML). It is the HTMX reveal
+// response AND the inline chat block on the no-JS chat page.
+templ chatSection(v chatView) {
+	<section id="chat-section" class="chat-dock chat-dock-open">
+		<h2 class="chat-heading">Ask about this video</h2>
+		<p class="chat-scope muted">Answers come only from this video's stored transcript — Tapir never fetches anything new here.</p>
+		@chatPanel(v)
+	</section>
+}
+
+// ChatPage is the no-JS full-page render of the chat: the whole summary followed
+// by the open chat dock, so a visitor without JS sees the same integrated view
+// (summary beside the conversation) that JS users get inline via the reveal.
+templ ChatPage(r store.SummaryRow, v chatView) {
+	@Layout("Tapir — " + displayTitle(r)) {
+		<article class="detail">
+			<p class="back"><a href="/">← Summaries</a></p>
+			<h1>{ displayTitle(r) }</h1>
+			@summaryBody(r)
+			@chatSection(v)
+		</article>
+	}
+}
+
+// chatPanel is the conversation + ask form, swapped in place on each answer
+// (HTMX targets #chat-panel, outerHTML). When no transcript is stored it shows the
+// honest "not available" state and no form (ADR-027: never a fetch). The prior
+// turns ride as hidden hq/ha fields so the ephemeral conversation survives the
+// round-trip without any persisted state.
+templ chatPanel(v chatView) {
+	<div id="chat-panel" class="chat-panel">
+		if !v.Available {
+			<p class="chat-unavailable muted">Chat isn't available for this video — its transcript isn't stored, and chat never fetches new captions. Summarize the video first to store its transcript.</p>
+		} else {
+			if len(v.History) > 0 {
+				<div class="chat-log">
+					for _, t := range v.History {
+						<div class="chat-turn chat-q"><p>{ t.Question }</p></div>
+						<div class="chat-turn chat-a"><p class="body">{ t.Answer }</p></div>
+					}
+				</div>
+			}
+			if v.Truncated {
+				<p class="chat-note muted">Working from a bounded portion of a long transcript — answers about the end of the video may be incomplete.</p>
+			}
+			if v.Error != "" {
+				<p class="chat-error" role="alert">{ v.Error }</p>
+			}
+			<form
+				class="chat-form"
+				method="post"
+				action={ chatURL(v.VideoID) }
+				hx-post={ string(chatURL(v.VideoID)) }
+				hx-target="#chat-panel"
+				hx-swap="outerHTML"
+			>
+				for _, t := range v.History {
+					<input type="hidden" name="hq" value={ t.Question }/>
+					<input type="hidden" name="ha" value={ t.Answer }/>
+				}
+				<label class="chat-model">
+					Model
+					<select name="model">
+						for _, m := range v.Models {
+							<option value={ m } selected?={ m == v.Selected }>{ m }</option>
+						}
+					</select>
+					<span class="chat-model-hint muted">Switch models to compare answers on the same transcript.</span>
+				</label>
+				<textarea name="question" rows="3" placeholder="Ask a question about this video…" required aria-label="Your question"></textarea>
+				<button type="submit" class="btn">Ask</button>
+				<span class="htmx-indicator chat-thinking">thinking…</span>
+			</form>
+		}
+	</div>
+}
+
 // RegisterPage is the explicit registration step (ADR-012): an authenticated Dex
 // subject with no tapir user picks a display name to create their account.
 // errMsg, when set, reports a validation problem on the prior POST.

test/acceptance/scenario_coverage_test.go

@@ -38,10 +38,11 @@ var scenarioCoverage = map[string]string{
 	"An authenticated user on the welcome page sees their way in and out": "TestWelcomeLoggedIn",
 
 	// connect_account.feature
-	"Connect a YouTube account":                          "TestCallbackExchangesAndRecordsConnection",
+	"Connect a YouTube account":                              "TestCallbackExchangesAndRecordsConnection",
-	"Connecting an account discovers videos immediately": "TestCallbackTriggersDiscovery",
+	"Connecting an account discovers videos immediately":     "TestCallbackTriggersDiscovery",
-	"Connecting summarizes my newest videos right away":  "TestNewestUnsummarizedVideoIDs",
+	"Connecting summarizes my best recent videos right away": "TestOnboardBurstVideoIDs",
-	"Tokens are never stored in the clear":               "TestCallbackExchangesAndRecordsConnection",
+	"The onboarding burst summarizes with a stronger model":  "TestBurstChainModelsLeadsWithOnboardModel",
+	"Tokens are never stored in the clear":                   "TestCallbackExchangesAndRecordsConnection",
 
 	// paste_url.feature
 	"Paste a valid YouTube URL":                              "TestPasteValidURLAddsAndRequests",
@@ -63,6 +64,19 @@ var scenarioCoverage = map[string]string{
 	"A returning subject passes straight through":                               "TestRegisteredSubjectPassesThrough",
 	"Deleting an account removes only my data and leaves other users untouched": "TestDeleteAccountWipesDataAndSecretsAndLogsOut",
 
+	// chat_transcript.feature (ADR-027)
+	"A summary view offers a deeper-dive into the video":         "TestChatEntryAffordanceOnSummaryView",
+	"The summary and the chat are on one page":                   "TestChatIntegratedWithSummaryOnSamePage",
+	"Ask a question answered from the stored transcript":         "TestChatAnswersFromStoredTranscriptWithoutAnyFetch",
+	"Chat never fetches captions or reaches YouTube":             "TestChatAnswersFromStoredTranscriptWithoutAnyFetch",
+	"A video with no stored transcript offers no chat":           "TestChatUnavailableWhenNoStoredTranscript",
+	"The default model is the summary's model and is switchable": "TestChatModelSwitcherDefaultAndSwitch",
+	"Switching models re-runs against the same transcript":       "TestChatModelSwitcherDefaultAndSwitch",
+	"The cloud model is hidden when cloud is disabled":           "TestChatCloudModelAbsentWhenDisabled",
+	"A long transcript is bounded and the chat says so":          "TestChatTruncationNoteShown",
+	"A multi-turn conversation is ephemeral":                     "TestChatMultiTurnHistoryIsEphemeral",
+	"Chat is reachable only from my own summary view":            "TestChatOnlyReachableForOwnVideo",
+
 	// 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",