triage-pr-review

Triage automated PR review feedback from CodeRabbit, Sourcery, Qodo, Greptile, and CodeQL.

5|1|Updated Aug 18, 2023
One-click install
npx skills add https://github.com/thpoll83/PolyKybdHost --skill triage-pr-review-thpoll83
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: triage-pr-review
Source: https://github.com/thpoll83/PolyKybdHost/tree/main/.claude/skills/triage-pr-review
Command: npx skills add https://github.com/thpoll83/PolyKybdHost --skill triage-pr-review-thpoll83

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Pull requests on PolyKybd repos attract five automated reviewers whose output is not uniformly trustworthy: bots post long walkthroughs without actually reviewing, rate-limit refusals render as green checks, and completed reviews can be pinned to a pre-push commit. This Skill determines which reviewers actually read the current head, verifies every finding against the real code, and drives the feedback to a single resolution reply. ## Core Features & Use Cases - Reviewer verification: Fetches PR comments and reviews via the GitHub MCP tools and distinguishes real reviews from rate-limit notices, summaries, refusals, and stale reviews pinned to old commits. - Finding validation: Reproduces or refutes each bot finding against the actual code before acting, fixing true findings and declining false ones with evidence. - Single consolidated reply: Posts one comment with a verdict table covering every finding, plus a section explaining declined items. - Use Case: After opening a PR and waiting for the bots, ask the assistant to check the PR for review feedback; it will report which of CodeRabbit, Sourcery, Qodo, Greptile, and CodeQL actually reviewed the head commit and address their verified findings. ## Quick Start Ask the assistant to check the pull request for review feedback and address what CodeRabbit, Sourcery, Qodo, Greptile, and CodeQL said.

Frequently Asked Questions about triage-pr-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check if a PR was actually reviewed by CodeRabbit or Sourcery?

Fetch the PR's reviews and comments, then confirm each review's commit_id equals the head sha and its body is not a refusal notice. A long walkthrough comment alone is not a review, and a green check run does not prove the current head was read.

How to triage automated code review feedback on a GitHub pull request?

Pull both get_comments and get_reviews, identify which bots genuinely reviewed the head commit, then reproduce or refute each finding against the actual code. Fix the true findings, decline the false ones with evidence, and post a single consolidated reply.

Why does a green Sourcery or Greptile check not mean the PR was reviewed?

Both bots can return a success check while no review happened: Sourcery submits a rate-limit notice as a review body, and Greptile's check run has been measured green over PRs it never reviewed. Only the review object and its commit_id are reliable evidence.

What does it mean when CodeRabbit posts no review object?

An empty get_reviews result can mean a clean pass, since CodeRabbit reports no actionable comments by editing its summary comment instead of creating a review. Read the summary comment body for its commits range and skipped-versus-clean wording before concluding anything.

When should I not use this PR review triage workflow?

Do not use it for CI failures; build job logs and the HIL failure diagnostic flow handle those. It is specifically for interpreting and acting on reviewer bot feedback on an open pull request.