Loop/goalCIhigh riskadvancedsafety C · 55GitHub

Ship OAuth MCP server, full regression

Execute the ship spec: write docs, pass all tests and macro checks, deploy or block with reasons, and deliver final acceptance criteria.

Prompt
→ Claude Code
/goal Execute specs/03-ship.md (read it and BRIEF.md first). Write README.md and SECURITY.md per the spec, then run the full macro regression and attempt deploy. Final turn must contain ALL of: (a) npx tsc --noEmit output with exit code 0 stated; (b) npm test full output with exit 0; (c) against a running wrangler dev, quoted curl -sS output of GET /.well-known/oauth-authorization-server AND GET /.well-known/oauth-protected-resource, both returning JSON; (d) quoted curl -i POST /mcp with no Authorization header showing HTTP 401 and a WWW-Authenticate header; (e) quoted curl -i POST /token with Content-Type: application/x-www-form-urlencoded and a bogus grant showing an HTTP 4xx OAuth JSON error body (not 415, not 500); (f) ls README.md SECURITY.md output plus the quoted markdown headings of both files; (g) quoted grep output showing README.md contains the literal strings GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRET, ALLOWED_HOSTED_DOMAIN, /callback, and all three Google OAuth scope URLs from BRIEF.md; (h) EITHER wrangler deploy output plus quoted curl of the deployed /.well-known/oauth-authorization-server returning JSON, OR the literal line "DEPLOY BLOCKED: <reason>" followed by the exact commands for me to run; (i) the literal heading "HUMAN ACCEPTANCE CHECKLIST" followed by the human-only steps from specs/03-ship.md verbatim. Stop after 10 turns and output a status report if not complete
claude-code

Source: zenf-star ↗graded C · 55/100 — how grades work (Safety Score v2) →

Similar loops

All ci loops →

Run every 15 minutes to classify CI failures, fix clear regressions in a worktree with verification, and escalate infrastructure or security test failures.

prompt
→ Claude Code
/loop 15m $ci-triage — update ci-sweeper-state.md. Classify failures first. Fix only clear regressions in a worktree with verifier. Max 3 attempts. Escalate infra and security test failures
/goalCIlow risk

Get the build green

Run the build, fix the first error, and repeat until npm run build exits 0, with a 10-turn cap.

prompt
→ Claude Code
/goal `npm run build` exits 0 — run the build, fix the first error, repeat until it succeeds; stop after 10 turns

Run ci-triage on failing CI, fix clear regressions in isolation, verify before push, and update state—up to 3 attempts per cluster.

prompt
→ Claude Code
/loop 15m Run ci-triage on failing CI for default branch and open PRs. Update ci-sweeper-state.md. Fix only clear regressions in an isolated worktree. Spawn loop-verifier before any push. Max 3 attempts per failure cluster