skills.rest
truongnat
truongnatCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

review

Reviews code changes against plans and produces a structured REVIEW.md with findings and recommendation.

Updated Jul 28, 2026
Search Tags:#diff analysis#code review#pull request#verification#quality gate#readiness#findings
One-click install
npx skills add https://github.com/truongnat/Restly --skill review-truongnat
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/truongnat/Restly/tree/main/.agents/skills/review
Command: npx skills add https://github.com/truongnat/Restly --skill review-truongnat

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? After executing code changes, teams need a disciplined, evidence-based check before marking work done or opening a PR. This Skill enforces a hard review contract that compares the diff and EXECUTION.md against PLAN.md and TASKS.md, producing a structured REVIEW.md instead of ad-hoc gut checks. ## Core Features & Use Cases - Evidence-based findings: Every finding requires severity, location, evidence, impact, and recommendation — no vague claims allowed. - Requirement coverage check: Maps each requirement or task to evidence showing whether the change actually covers it. - Security and risk gates: Forces explicit checks when changes touch auth, permissions, secrets, files, network, or databases, and blocks a Ready verdict while artifact validators fail. - Use Case: After finishing a feature branch, run the review to get a REVIEW.md with findings by severity, testing gaps, residual risks, and a clear Ready / Needs fix / Blocked recommendation before opening the PR. ## Quick Start Use the review skill to evaluate the current diff against PLAN.md and EXECUTION.md and produce a REVIEW.md with findings and a readiness recommendation.

Frequently Asked Questions about review

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

FAQPage Schema
How do I review code changes before creating a pull request?▼

Run the review after execution: it compares the diff and EXECUTION.md against PLAN.md and TASKS.md, then writes a REVIEW.md with findings, verification evidence, and a readiness recommendation. It does not auto-fix code.

What makes a code review finding evidence-based?▼

Each finding must include an ID, severity (Critical/High/Medium/Low/Info), category, exact file location, concrete evidence such as a diff context or line, impact, and a recommendation. Vague statements without evidence are explicitly rejected.

Can the review skill fix the issues it finds?▼

No. The contract explicitly forbids auto-fixing code when the user only requested a review. If fixes are needed, the workflow returns to planning or execution with updated PLAN.md or TASKS.md.

When does the review block a Ready recommendation?▼

Ready is blocked while validate_artifacts.py or lint_artifacts.py fails on the active session, when verification evidence is missing, or when unaddressed security and data risks exist in changes touching auth, permissions, secrets, files, network, or databases.

What are the limitations of an automated code review?▼

It does not replace full QA or a deep security audit, and it cannot claim a full review when inputs like the diff are incomplete. Pre-existing issues are documented as informational findings rather than attributed to the current change.

Related Skills

Explore More Software Engineering Skills→
microsoftmicrosoft
1.4k

rpi-review

Compare RPI planning and implementation evidence and record severity-graded review findings.

Official
Advanced
thesyntaxninjathesyntaxninja
2

self-review

Review git diffs for deletions, simplifications, and pattern mismatches.

Community
Intermediate
EmasoftEmasoft
13

amoa-verification-patterns

Generate evidence-based JSON verification reports for implemented code and systems.

Community
Advanced
PiLastDigitPiLastDigit
614

TRIP-review

Reviews code changes against a project checklist and generates structured review reports.

Community
Intermediate