메뉴
Investigate and fix the latest failing CI workflow
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Open the Claude Session Dashboard in the browser. Starts it first if not already running. Checks for and installs updates before starting.
Analyze ~/.claude session files to find unused agent skills and reduce context token waste. Scans subagent JSONL files for injected skills and cross-references with agent responses.
Full SDLC pipeline: design → plan → implement → review → test → ship. Use this skill when the user wants to BUILD A NEW FEATURE that spans multiple files across scanner, server functions, and UI layers — anything requiring architecture design, an implementation plan, and coordinated changes across a vertical slice. Trigger on: /feature command, 'build/implement/add [feature]' requests involving new server functions + UI components, multi-step feature work needing TDD and code review. Do NOT trigger for: bug fixes, refactors, small UI tweaks, single-file changes, CI fixes, test runs, or PR reviews.
Fix a GitHub issue end-to-end
Investigate a URL or page using Playwright browser
Analyze a change request, ask clarifying questions, and create a well-structured GitHub issue with acceptance criteria. Use this skill whenever the user wants to create an issue, write a ticket, log a bug, propose a feature request, or capture requirements — even if they say "let's track this" or "add this to the backlog".
| name | fix-ci |
| description | Investigate and fix the latest failing CI workflow |
| user-invocable | true |
| argument-hint | [run-id] |
Investigate the latest CI failure, reproduce locally, and fix it.
# Get latest (or specific) workflow run
gh run list --limit 5
gh run view <run-id>
If $ARGUMENTS.run-id is provided, use that. Otherwise pick the latest failed run.
gh run view <run-id> --log-failedRead the logs and classify the failure:
| Type | Indicators |
|---|---|
| typecheck | error TS, tsc --noEmit |
| lint | eslint, warning/error with rule names |
| unit test | vitest, FAIL, expect( |
| e2e test | playwright, spec.ts, timeout, locator |
| build | npm run build, vite build, rollup |
| dependency | npm ci, ERESOLVE, peer dep |
Extract the exact error message, file path, and line number.
Use superpowers:systematic-debugging to find root cause before attempting any fix.
Run the failing CI step locally from apps/web/:
| CI Step | Local Command |
|---|---|
| typecheck | npm run typecheck |
| lint | npm run lint |
| unit test | npm run test |
| e2e test | npm run test:e2e |
| build | npm run build |
| install | npm ci |
If reproducing a specific test:
npx vitest run <test-file>npx playwright test <spec-file>If the failure is an E2E test or a visual/runtime bug:
lsof -i :3000mcp__playwright__browser_navigatemcp__playwright__browser_take_screenshotmcp__playwright__browser_console_messagesmcp__playwright__browser_network_requestsmcp__playwright__browser_snapshotRun the exact failing step again locally:
cd apps/web
# Run the specific failing command
Then run the full quality suite to ensure no regressions:
cd apps/web
npm run typecheck
npm run lint
npm run test
npm run build
Output a summary:
## CI Fix Summary
- Run: #<run-id> (<branch>)
- Failed job: <job-name>
- Root cause: <description>
- Fix: <what was changed>
- Local verification: PASS / FAIL
mcp__playwright__browser_close