skills/review-pr/SKILL.md
Review a pull request — interactive PR selection via `gh`, 4-dimension code review (hallucinated code, architecture, performance, test validity), then post severity-graded comments with fix suggestions on the PR. Not for spec-based review — use `review` instead.
npx skillsauth add laitszkin/apollo-toolkit review-prInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
gh pr comment posts completed successfully| Level | When to assign | |-------|----------------| | P0 — Critical | Main functionality broken; business logic contradicts design intent | | P1 — Major | Secondary functionality broken | | P2 — Minor | Edge functionality broken | | P3 — Trivial | Rare boundary conditions cause incorrect behavior |
Examine every changed file through all four lenses. For each finding record: file, line range, severity, description, and suggested fix. Cross-reference dimensions — one change may have findings in multiple.
When reviewing, watch for:
If the user provided a PR number or URL, confirm it with gh pr view. Otherwise list open PRs, present a table, and let the user pick.
Fetch PR title, body, files, diff, and existing comments. Note base and head branches — they frame scope.
Read every changed file. Group findings by severity (P0 → P3).
Verify completeness: cross-reference the list of changed files against your findings. For any file with no findings, confirm to the user it was reviewed and deemed clean before proceeding.
Show the user the full finding list grouped by severity with a summary table. Ask explicitly whether to post — wait for confirmation.
Post one gh pr comment per finding. Each comment follows this format:
**P# — Dimension Name**
**File:** `path/file.ts:L42`
**Issue:** clear description
**Suggested Fix:** actionable suggestion (with code if applicable)
Verify posting: After all posts, check gh pr view --json comments to confirm every finding has a matching comment. Compare actual post count to expected. Then summarize: Posted N findings (P0: x, P1: y, P2: z, P3: w) on PR #N.
gh CLI installed and authenticated (gh auth status)origin, or user provides --repo owner/namereferences/gh-pr-cheatsheet.md for exact gh pr flagsdevelopment
Read a user-specified PDF that marks the week's key financial events, deeply research each marked event with current sources, capture any additional breaking financial developments, and produce a concise Chinese-capable PDF briefing that explains what happened and why it matters.
documentation
Generate long-form videos (more than 10 minutes) by following user instructions and invoking related skills only when needed (`openai-text-to-image-storyboard`, `docs-to-voice`, `remotion-best-practices`). For text inputs, extract a complete long-form story arc, generate fresh storyboard images (no reuse of previously generated pictures), and render a 16:9 animated long-form video.
tools
協助完成自動化版本發佈。同步文檔、更新版本號、推送 tag 並建立 GitHub Release。
development
Incrementally refresh the architecture atlas when the project diagram drifts from actual code. Measures drift before updating to determine scope, then updates the base atlas and re-renders HTML.