skills.rest
alessioroberto82
alessioroberto82COMMUNITY
Explore all skills by this author →
GitHub

pr-review

Reviews pull requests with parallel agents against project standards and security rules.

1|Updated Feb 19, 2026
Search Tags:#automation#static analysis#code review#pull request#security#github#multi-agent
One-click install
npx skills add https://github.com/alessioroberto82/claude-plugin-circle --skill pr-review-alessioroberto82
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: pr-review
Source: https://github.com/alessioroberto82/claude-plugin-circle/tree/main/plugins/circle/skills/pr-review
Command: npx skills add https://github.com/alessioroberto82/claude-plugin-circle --skill pr-review-alessioroberto82

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Manual pull request review is slow and inconsistent, often missing violations of project conventions or real security issues buried in large diffs. This Skill automates multi-agent PR review grounded in your repository's own AGENTS.md, CLAUDE.md, and language best practices, posting only high-confidence, citation-backed findings. ## Core Features & Use Cases - Multi-Agent Parallel Review: Runs a standards/bugs agent and a security agent (CWE/OWASP) in parallel, plus optional platform-specific review skills discovered automatically. - Project Context Awareness: Reads root and nested AGENTS.md/CLAUDE.md files, .agents/ and .claude/ docs, ADRs, and PR descriptions so intentional design decisions are not flagged as regressions. - Strict Quality Gates: Filters findings by confidence thresholds (90, or 75 for foundational files), requires citations, and discards known false positives before posting a GitHub comment via gh. - Use Case: A developer opens a PR touching API code; the Skill detects the language, loads relevant best-practice skills, reviews the diff against repo standards, and posts a single consolidated review comment with file/line links. ## Quick Start Ask the assistant to review pull request number 123 using the pr-review skill and post the findings as a GitHub comment.

Frequently Asked Questions about pr-review

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

FAQPage Schema
How do I review a GitHub pull request with AI agents?▼

Provide a PR number, URL, or branch name and the skill fetches the diff and metadata via the gh CLI, then runs parallel review agents for standards, bugs, and security. Findings are filtered by confidence and posted as a single PR comment.

How does automated PR review use AGENTS.md or CLAUDE.md files?▼

The review reads root and nested AGENTS.md and CLAUDE.md files plus .agents/ and .claude/ docs, treating them as the binding project standards. Every flagged issue must cite the exact rule text from those files, and uncited findings are discarded.

Can PR review avoid flagging intentional design changes?▼

Yes. The skill reads the PR description and any ADR files in the diff, so behavioral changes justified by a documented design decision are treated as intentional. Confidence is capped at 25 for findings that contradict an ADR or a referenced cross-platform implementation.

What security issues does automated code review detect?▼

The security agent scans the diff for injection, authentication gaps, hardcoded secrets, weak cryptography, and data exposure. Each finding must cite a CWE or OWASP reference such as CWE-79 or CWE-798, otherwise it is discarded.

Why are some review findings not posted to the GitHub comment?▼

Findings below the confidence threshold (90, or 75 for foundational files), lacking citations, or matching the false positive guide are filtered out. Near-miss findings are recorded only in a local summary file, never posted publicly.

Related Skills

Explore More Software Engineering Skills→
malinskibeniaminmalinskibeniamin
3

setup-matt-pocock-skills

Scaffold repository starter files and triage-label mappings for issue tracking.

Community
Intermediate
jlui17jlui17

pr-review

Reviews pull requests and diffs against repository conventions with verified findings.

Community
Intermediate
vtgiang-dotcomvtgiang-dotcom
3

requesting-code-review

Verify staged Git changes with security scans and baseline checks before commits.

Community
Intermediate
EngineerWithAIEngineerWithAI
4

sast-configuration

Configure Semgrep, SonarQube, and CodeQL for automated vulnerability detection.

Official
Intermediate