2026-04-14 - 2026-10-14
Overview
55 Releases published by 2 users
Published
v0.22.1
Published
v0.22.0
Published
v0.21.0
Published
v0.20.1
Published
v0.19.7
Published
v0.20.0
Published
v0.19.6
Published
v0.19.5
Published
v0.19.4
Published
v0.19.3
Published
v0.19.2
Published
v0.19.1
Published
v0.19.0
Published
v0.18.2
Published
v0.18.1
Published
v0.18.0
Published
v0.17.1
Published
v0.17.0
Published
v0.16.0
Published
v0.15.1
Published
v0.15.0
Published
v0.14.0
Published
v0.13.4
Published
v0.13.3
Published
v0.13.2
Published
v0.13.1
Published
v0.13.0
Published
v0.12.1
Published
v0.12.0
Published
v0.11.0
Published
v0.10.2
Published
v0.10.1
Published
v0.10.0
Published
v0.9.0
Published
v0.8.5
Published
v0.8.4
Published
v0.8.3
Published
v0.8.2
Published
v0.8.1
Published
v0.8.0
Published
v0.7.0
Published
v0.6.3
Published
v0.6.2
Published
v0.6.1
Published
v0.6.0
Published
v0.5.0
Published
v0.4.2
Published
v0.4.1
Published
v0.4.0
Published
v0.3.3
Published
v0.3.2
Published
v0.3.1
Published
v0.3.0
Published
v0.2.0
Published
v0.1.0
4 Pull requests merged by 1 user
Merged
#52 feat(hooks): mcp-destructive-guard asks before irreversible MCP verbs
Merged
#21 feat(own-the-task): add delegation-ownership skill
Merged
#1 feat: add experiment-spec skill for research projects
Merged
#2 feat(skills): add dream skill
34 Issues closed from 1 user
Closed
#43 GITEA_MCP_TOKEN is stale — it 401s against every gitea-mcp endpoint
Closed
#51 pr-gatekeeper exists only in ~/.claude/agents on koala — the estate's most effective control is unversioned
Closed
#44 hooks/pre-task-ritual.sh has no test
Closed
#33 Wire gitea-mcp into koala's local .mcp.json (Tailscale Serve) -- needs a NodePort first
Closed
#41 Re-add an Antigravity target to install.sh (previously removed as "unused")
Closed
#19 Execution bar for #9-18: pressure-test every new/edited skill, prefer small tools over prose
Closed
#10 New skill: traceability — forward requirement/decision graph (spec→ADR→commit→attestation)
Closed
#11 New skill: lineage — backward decision archaeology
Closed
#9 New skill: adr — codify cad's ADR template + productive-tensions pattern
Closed
#20 Add Python conventions to tdd/clean-code/solid/code-review per-language reference files
Closed
#18 Generalize tdd skill beyond Go — add Python, Rust, TypeScript, Java
Closed
#36 debug: no intake gate, a one-command rule that breaks on infra, and a session_log payload missing session_id
Closed
#38 doctor cannot see a release the host has not fetched — clean at 0.12.1 while main is v0.13.0
Closed
#17 Follow-up: adapt superpowers-* planning/execution cluster to this harness's actual Agent/Workflow tools
Closed
#5 Should SKILL.md define explicit escalation / failure conditions?
Closed
#12 Edit spec-driven-dev + feature-spec: EARS-lite acceptance criteria + inline context snapshot
Closed
#14 Merge 2 techniques from superpowers-systematic-debugging into debug, then delete it
Closed
#16 Cross-link code-review trio + grill-me→stage-gate-review; strip false CLAUDE.md citation in receiving-code-review
Closed
#29 Declare a surfaces: field per skill; gate in check, filter install.sh, generate SKILLS_INDEX.md
Closed
#34 task check goes red on a dev checkout whose tags are stale, not on a real fault
Closed
#27 install.sh assumes one source of skills — a second source can only be served by displacing the first
Closed
#30 doctor: a Claude Code host on a stale plugin version reports healthy
Closed
#7 task install destroys/skips skill symlinks — use install.sh instead
Closed
#13 Delete superpowers-test-driven-development — near-verbatim duplicate of tdd
Closed
#26 web-shot: SKILL.md documents a shot.sh path that does not exist
Closed
#15 SKILLS_INDEX.md missing experiment-spec and dream rows
Closed
#25 secret-guard: env -u VAR cmd is treated as an env dump and blocked
Closed
#22 Clean up stale branches (and decide on a minimal CI gate)
Closed
#31 cad-babysit: a focused human-in-the-loop skill for working CAD's stuck queue
Closed
#24 Roll the plugin out to flamingo and iguana — both currently run with no skills and no secret-guard
Closed
#23 install.sh: per-repo wiring scatters absolute symlinks that rot when the checkout moves
Closed
#8 fix(ci): release.yml tag job broken since 2026-06-22
Closed
#4 feat: telos-load skill — harness-agnostic TELOS session context loading
Closed
#3 feat: regulatory-risk-assessment skill — CAD compliance gate
51 Issues created by 1 user
Opened
#3 feat: regulatory-risk-assessment skill — CAD compliance gate
Opened
#4 feat: telos-load skill — harness-agnostic TELOS session context loading
Opened
#5 Should SKILL.md define explicit escalation / failure conditions?
Opened
#6 feat: gitea-rest + brain-rest skills — default to direct REST API on CLI to bypass MCP-relay Cloudflare WAF blocks
Opened
#7 task install destroys/skips skill symlinks — use install.sh instead
Opened
#8 fix(ci): release.yml tag job broken since 2026-06-22
Opened
#9 New skill: adr — codify cad's ADR template + productive-tensions pattern
Opened
#10 New skill: traceability — forward requirement/decision graph (spec→ADR→commit→attestation)
Opened
#11 New skill: lineage — backward decision archaeology
Opened
#12 Edit spec-driven-dev + feature-spec: EARS-lite acceptance criteria + inline context snapshot
Opened
#13 Delete superpowers-test-driven-development — near-verbatim duplicate of tdd
Opened
#14 Merge 2 techniques from superpowers-systematic-debugging into debug, then delete it
Opened
#15 SKILLS_INDEX.md missing experiment-spec and dream rows
Opened
#16 Cross-link code-review trio + grill-me→stage-gate-review; strip false CLAUDE.md citation in receiving-code-review
Opened
#17 Follow-up: adapt superpowers-* planning/execution cluster to this harness's actual Agent/Workflow tools
Opened
#18 Generalize tdd skill beyond Go — add Python, Rust, TypeScript, Java
Opened
#19 Execution bar for #9-18: pressure-test every new/edited skill, prefer small tools over prose
Opened
#20 Add Python conventions to tdd/clean-code/solid/code-review per-language reference files
Opened
#22 Clean up stale branches (and decide on a minimal CI gate)
Opened
#23 install.sh: per-repo wiring scatters absolute symlinks that rot when the checkout moves
Opened
#24 Roll the plugin out to flamingo and iguana — both currently run with no skills and no secret-guard
Opened
#25 secret-guard: env -u VAR cmd is treated as an env dump and blocked
Opened
#26 web-shot: SKILL.md documents a shot.sh path that does not exist
Opened
#27 install.sh assumes one source of skills — a second source can only be served by displacing the first
Opened
#28 Three skills exist only on claude.ai — unmirrored writers outside git
Opened
#29 Declare a surfaces: field per skill; gate in check, filter install.sh, generate SKILLS_INDEX.md
Opened
#30 doctor: a Claude Code host on a stale plugin version reports healthy
Opened
#31 cad-babysit: a focused human-in-the-loop skill for working CAD's stuck queue
Opened
#33 Wire gitea-mcp into koala's local .mcp.json (Tailscale Serve) -- needs a NodePort first
Opened
#32 Wire cad-mcp into koala's local .mcp.json (Tailscale Serve, next free port)
Opened
#34 task check goes red on a dev checkout whose tags are stale, not on a real fault
Opened
#35 flamingo: wire the parked Anthropic set as an extra source, and rescue gotth-dev from it
Opened
#36 debug: no intake gate, a one-command rule that breaks on infra, and a session_log payload missing session_id
Opened
#37 Implement the escalation gate decided in #5: ## Escalation required for acting skills
Opened
#38 doctor cannot see a release the host has not fetched — clean at 0.12.1 while main is v0.13.0
Opened
#39 session_log asks for a session_id no agent can obtain — two runs declined to log rather than invent one
Opened
#40 doctor blames the Claude Code plugin for drift when run from a dev clone
Opened
#41 Re-add an Antigravity target to install.sh (previously removed as "unused")
Opened
#42 CI runner has no jq, so Antigravity's merge behaviour is now skipped rather than tested
Opened
#43 GITEA_MCP_TOKEN is stale — it 401s against every gitea-mcp endpoint
Opened
#44 hooks/pre-task-ritual.sh has no test
Opened
#45 task doctor: assert build-Go >= GOROOT for every Go analysis tool the gate invokes
Opened
#46 Every release leaves main's HEAD with zero CI runs, so CIHeadUncovered fires forever and can never clear
Opened
#47 Considered and declined: renaming the plugin MCP servers to brain-local / gitea-local / infra-local
Opened
#48 Ship the MCP-reminder counter-hook from the plugin — it exists on koala only, and is mitigation not enforcement
Opened
#49 Add a committed loop-autonomy contract + fingerprint dedupe to the CAD loop skills
Opened
#50 adr skill: the template's unquoted adr: NNNN is a YAML octal trap that bites at 0008, 0009, 0018…
Opened
#51 pr-gatekeeper exists only in ~/.claude/agents on koala — the estate's most effective control is unversioned
Opened
#53 doctor: third-party skills in ~/.agents/skills fail SessionStart hook as "unmanaged"
Opened
#54 Release-gate hook: creating a tag (git tag, tag_create, tbd_ship) must run the repo's own release gate
Opened
#55 plugin brain MCP on koala rejects its Authorization header mid-session (2026-10-11)