review

Review pull request diffs against CONTEXT.md principles with severity-tagged findings.

Updated Mar 2, 2022
One-click install
npx skills add https://github.com/ysuurme/azure_hello_world --skill review-ysuurme
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/ysuurme/azure_hello_world/tree/main/.agents/skills/review
Command: npx skills add https://github.com/ysuurme/azure_hello_world --skill review-ysuurme

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents architectural drift and defects from reaching main by running a structured PR review cycle with automated fixes when findings are actionable.

Core Features & Use Cases

  • Senior Architect PR review: Reviews PR diffs against CONTEXT.md principles and review heuristics, producing severity-tagged findings (BLOCK/WARN/INFO).
  • Auto-fix loop for actionable issues: Applies fixes in an isolated worktree, verifies with tests, and re-evaluates the specific finding until it is resolved or escalated.
  • HITL escalation gate: Stops after 3 iterations per finding, blocks merge for unresolved BLOCK/WARN issues, and escalates with a human-in-the-loop comment and label.
  • Verification and compaction: Re-checks findings against the current diff and consolidates related findings into a clear review outcome for ship readiness.

Quick Start

Use this Skill when you have an open pull request ready for review and want an enforced, standard-aware review that can automatically fix issues or block with HITL escalation if needed.

Frequently Asked Questions about review

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

FAQPage Schema
How do I automate pull request reviews to prevent architectural drift?

Automated pull request reviews prevent architectural drift by validating PR diffs against architecture standards using severity-tagged findings. The review process applies auto-fix loops in an isolated worktree and blocks merges when unresolved violations are detected.

What happens when a PR review finds a BLOCK or WARN issue?

When a PR review finds a BLOCK or WARN issue, an auto-fix loop attempts automated correction in an isolated worktree with test verification. If the finding remains unresolved after three iterations, it triggers a human-in-the-loop escalation.

How does continuous integration handle unresolved architectural violations?

Continuous integration handles unresolved architectural violations by blocking the merge gate and escalating with a human-in-the-loop comment and label. This severity-gated mechanism ensures defects never reach the main branch without manual review.

Can I auto-fix code review findings without merging broken code?

You can auto-fix code review findings without merging broken code by applying corrections in an isolated worktree. The system verifies fixes with tests and re-evaluates the specific finding before allowing the pull request to become PR-ready.

When do I need a human-in-the-loop escalation gate for code reviews?

A human-in-the-loop escalation gate for code reviews is needed when automated auto-fix loops fail to resolve a BLOCK or WARN finding within three iterations. It stops the automated process and blocks the merge until manual intervention occurs.

Does the PR review process support severity-gated merge blocking?

The PR review process supports severity-gated merge blocking by tagging findings as BLOCK, WARN, or INFO. It blocks the downstream ship gate for any unresolved BLOCK or WARN issues and compacts related findings into a clear review outcome.