Repository navigation
chore(docs): update AGENTS.md to reflect recent changes - #112
Conversation
AGENTS.md's PostgreSQL setup steps pointed at compose.dev.yml, a file that has never existed in this repo; the actual file is .docker/compose.yml. Also document the one-time postgres-data volume creation the file's external:true declaration requires, since docker compose up otherwise fails on a fresh environment. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KjpmZuodwjnvc8Q93y21ZG
![pullfrog[bot]](https://avatars.githubusercontent.com/in/1768019?s=60&v=4){kind=small}
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes
Reviewed the single-commit doc fix to AGENTS.md's PostgreSQL setup instructions.
- Fixed stale compose file path:
docker compose -f compose.dev.yml ...→docker compose -f .docker/compose.yml .... Confirmed.docker/compose.ymlis the only compose file in the repo andcompose.dev.ymlhas never existed. - Documented the
postgres-dataexternal-volume prerequisite: added adocker volume create postgres-datastep before the firstup. Confirmed in.docker/compose.ymlthat thepostgres-datavolume is declaredexternal: true, so this step is factually necessary on a fresh environment.
Grepped the repo for any other compose.dev.yml/compose.yml references — none remain stale outside this diff. Both claims check out against current repo state; this is a clean, minimal, and accurate documentation fix.
Claude Sonnet ｜ 𝕏
…p works Compose rejected the whole project (even with only `postgres` selected) because the `redis` service referenced an undeclared `redis-data` volume, and the `ring` service's `env_file` pointed at `src/.env.development.local`, which AGENTS.md never tells anyone to create. Declare the volume and mark that env file optional so `docker compose -f .docker/compose.yml up postgres -d` validates on a fresh checkout with no local env overrides in place. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KjpmZuodwjnvc8Q93y21ZG
◈ PR LensNote The title starts with
|
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes
Reviewed the second commit (6e60082), which fixes the two docker compose validation failures that Greptile flagged against the previous commit's documented PostgreSQL startup.
- Declared the
redis-datavolume: addedredis-data:under the top-levelvolumes:in.docker/compose.yml, matching theredisservice's existingredis-data:/datamount. Without this, Compose rejected the whole project (service redis refers to undefined volume redis-data) even when onlypostgreswas selected. - Made the dev-local env file optional: converted the
../src/.env.development.localenv_fileentry to the long form withrequired: false, sodocker compose ... up postgres -dno longer fails on a fresh checkout with no local override files present..env.development.localis a legitimate gitignored override file (.gitignore:28), so this isn't a stray reference — it's still loaded when present.
I independently confirmed both root causes against the current .docker/compose.yml: the redis service block mounts redis-data and was previously undeclared at the top level, and src/.env.development.local doesn't exist anywhere in this checkout (only src/.env does), so required: false is the correct fix rather than a workaround. This directly resolves the "not safe to merge" blocker from the prior review round.
Claude Sonnet ｜ 𝕏
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
|
Status: waiting on CI infra, not this PR. All content-relevant checks are green: Greptile (confidence 5/5, "No outstanding finding blocks merging"), pullfrog, CodeQL, Socket Security (both), and GitGuardian. Both Greptile P1 findings from the earlier round (undeclared The two remaining required jobs — I'll keep watching this PR and will act on anything that changes (CI finally running, new findings, etc.). Generated by Claude Code |
|
Update: attempt 1 of the Everything content-relevant remains green (Greptile 5/5, pullfrog, CodeQL, Socket, GitGuardian). I'll keep watching and will re-run/merge readiness-check once Lint/Build actually execute. Generated by Claude Code |
2 participants
Summary
Weekly AGENTS.md maintenance review (window: 2026-09-21 to 2026-09-28). Only one PR merged into
mainin that window — #109, itself last week's AGENTS.md maintenance PR — plus one direct commit (334c946, "Addpullfrog.ymlworkflow") that added a manualworkflow_dispatch-only GitHub Action. That workflow doesn't run on push/PR, add dependencies, or change any build/lint/test command, so it needs no AGENTS.md coverage.While verifying AGENTS.md against the live repo, I found a real, still-unfixed gap: the "Cursor Cloud specific instructions → Infrastructure (PostgreSQL)" section references
compose.dev.yml, a file that has never existed in this repo — the actual file is.docker/compose.yml. This was previously identified and a fix attempted in PR #106, but that PR was closed unmerged (it went stale/conflicted when #109 touched the same file in the same window) before landing.I also re-verified against
.docker/compose.ymlthat itspostgres-datavolume is declaredexternal: true, sodocker compose ... up postgres -dfails on a fresh environment unless the volume is created first — nothing else in the repo creates it.Everything else in
AGENTS.mdand.github/copilot-instructions.mdwas cross-checked againstpackage.jsonscripts,src/lib/common/limits.ts, and theINTERNAL_API_STARYL_TOKEN/DISCORD_PUBLIC_KEY/DISCORD_TOKENreferences insrc/lib, and remains accurate.Changes
AGENTS.md: fixcompose.dev.yml→.docker/compose.yml, and document thepostgres-dataexternal-volume prerequisite (docker volume create postgres-data) as a one-time step before firstup.🤖 Generated with Claude Code
https://claude.ai/code/session_01KjpmZuodwjnvc8Q93y21ZG
Generated by Claude Code
No outstanding finding blocks merging.
Reviews (2) · Last reviewed commit: "fix(docker): make compose.yml valid so t..."