pr-review

Audit pull request context, diff, and touched files for a pass/fail verdict.

1|Updated Jan 6, 2026
One-click install
npx skills add https://github.com/Levezze/devkit --skill pr-review-levezze
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pr-review
Source: https://github.com/Levezze/devkit/tree/main/skills/pr-review
Command: npx skills add https://github.com/Levezze/devkit --skill pr-review-levezze

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prevents merges of PRs from ending the development cycle with unreviewed, potentially harmful, or policy-violating changes by enforcing a diligent, structured, evidence-based review before integration.

Core Features & Use Cases

  • Read-only PR review by default: Presents findings to the user and never applies fixes or delegates implementation work automatically, avoiding accidental mutation of the branch.
  • Evidence-first auditing workflow: Reads PR metadata, issue/PRD context, and the full diff and touched files, then synthesizes a final verdict from streamed audit artifacts.
  • Safety guardrails & anti-pattern detection: Checks for patch/wrapper/hand-wave/hidden-mock/architecture-conformance/coverage smells and fails on critical or unresolved issues.
  • Resumable review artifacts: Streams outputs to durable /tmp/pr-review-<pr#>-<TS>-*.md files and supports resuming interrupted runs using the existing on-disk state.

Quick Start

Invoke pr-review before merging by telling your AI to run: pr-review for PR number N.

Frequently Asked Questions about pr-review

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

FAQPage Schema
How do I automate a read-only pull request review before merging?

Automated read-only pull request review audits PR metadata, issue context, and full diffs to synthesize a pass/fail verdict without applying fixes or mutating the branch.

What is a fail-fast quality gate for code review?

A fail-fast quality gate enforces a structured end-of-cycle review before merge by detecting patch smells, wrapper patterns, and coverage gaps to block harmful policy-violating changes.

How does resumable code audit work if a review gets interrupted?

Resumable code audit streams findings to durable /tmp files with consistent filenames and timestamps, allowing interrupted runs to resume using existing on-disk state.

Can I detect architecture conformance and mocking anti-patterns in a PR?

Detecting architecture conformance and mocking anti-patterns in a PR checks the full diff and touched files for hidden mocks, TODOs, wrappers, and policy compliance violations.

Does the PR review process support multi-PR merges?

The PR review process supports multi-PR merges by running repeatedly before integration when the author intends to combine changes, auditing each PR context thoroughly.

What limitations exist for automated PR policy compliance checks?

Automated PR policy compliance checks operate as read-only by default without applying fixes, require full diff context for evidence-based synthesis, and fail strictly on critical unresolved issues.