Agent skill
pr-review-helper
Gather PR review context: all comments (including review suggestions), diff context, and CI status; download GitHub Actions logs on failures. Prefer GitHub MCP for reads, fall back to gh CLI when needed.
Install this agent skill to your Project
npx add-skill https://github.com/hawkingrei/tidb-dev-prompt/tree/main/pr-review-helper
SKILL.md
pr-review-helper
Collect baseline information for PR review with minimal back-and-forth. Prefer GitHub MCP for all reads; if MCP lacks data or is unavailable, fall back to gh CLI.
Required Inputs
- Repository owner/name
- PR number
If any are missing, ask the user once. Do not guess.
Workflow (Prefer GitHub MCP)
-
PR overview
- Use
mcp__github__pull_request_readwithmethod: getto capture title, author, base/head branches, and head SHA.
- Use
-
All comments and review suggestions
- Review comment threads:
mcp__github__pull_request_readwithmethod: get_review_comments.- Extract: file path, line, author, body.
- If a suggestion block exists, capture the suggested patch and map it to the diff hunk.
- Review summary:
mcp__github__pull_request_readwithmethod: get_reviews.- Capture approvals, change requests, and top-level review bodies.
- PR/issue comments:
mcp__github__pull_request_readwithmethod: get_comments.- Include non-review discussion.
- Review comment threads:
-
Diff context
- Use
mcp__github__pull_request_readwithmethod: get_difffor full diff. - If the diff is large, also request
method: get_filesto scope by file before drilling into relevant hunks.
- Use
-
CI status
- Use
mcp__github__pull_request_readwithmethod: get_statusfor the head SHA. - Identify failing checks. If failures appear to be GitHub Actions, proceed to logs.
- Use
-
GitHub Actions logs (when failures exist)
- If MCP does not provide sufficient log details, fall back to
ghCLI. - Suggested flow:
gh pr checks <PR>to list failing runs and run IDs.gh run view <run-id> --log-failedto extract the failure reason.- If logs are truncated or need full context:
gh run download <run-id>then inspect the downloaded logs.
- If MCP does not provide sufficient log details, fall back to
Fallback Workflow (gh CLI Only)
Use this when MCP is unavailable or incomplete.
-
PR overview
gh pr view <PR> --json title,author,baseRefName,headRefName,headRefOid
-
Comments and review threads
gh pr view <PR> --json comments,reviews,reviewThreads- Parse
reviewThreadsfor file/line and suggestion blocks.
-
Diff
gh pr diff <PR>
-
CI status and logs
gh pr checks <PR>gh run view <run-id> --log-failedgh run download <run-id>if full logs are needed
Output Expectations
Provide a concise summary with:
- Review comments grouped by file/line, including suggestion patches if present
- Non-review discussion comments that may affect changes
- CI status: pass/fail, failing jobs, and key error messages
- Clear action items for the PR author
Do not post new comments, re-run CI, or update the PR unless the user explicitly asks.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
tidb-optimizer-bugfix
Fix TiDB optimizer bugs with minimal diffs, hypothesis-driven validation, and regression tests aligned with tidb-test-guidelines. Use when reproducing, fixing, and validating planner/optimizer behavior bugs.
plan-replayer-testing
Expertise in adding new test cases for the TiDB plan replayer. Use when the user provides a plan replayer zip file and wants to create a new test.
tidb-profiler-analyzer
Processes and analyzes TiDB profiler zip packages (CPU or heap) for components like TiDB, TiKV, PD, or TiFlash. It unzips, aggregates the profiler data, and uses `go tool pprof` to report on the most time-consuming functions (CPU) or highest memory allocations (heap).
tidb-doc-finder
TiDB-specific documentation lookup guided by this repo's llms.txt hub: read llms.txt, pick the best linked source (tidb-dev-guide llms-full.txt, TiDB user guide llms.txt, or Uber Go style guide), fetch it, then search within it to answer with precise references. Use when asked to find docs, research, or provide doc-grounded answers about TiDB development or usage.
context-management
Use for projects, tasks, or code that involve context engineering and context management for AI agents: keep prompt prefixes stable for KV-cache, use append-only context, prefer tool masking over tool removal, offload large observations into filesystem memory, recite goals/todos to control attention, preserve errors for recovery, and avoid few-shot pattern lock-in. Use when building or debugging agent loops, prompt/context schemas, memory strategies, or tool-availability policies.
edit-article
Edit and improve articles by restructuring sections, improving clarity, and tightening prose. Use when user wants to edit, revise, or improve an article draft.
Didn't find tool you were looking for?