skills.rest
psaab
psaabCOMMUNITY
Explore all skills by this author →
GitHub

review-pr

Coordinate multiple AI agents to adversarially review GitHub pull requests.

8|2|Updated Feb 8, 2026
Search Tags:#automation#code review#github#adversarial#network-expert#xpf
One-click install
npx skills add https://github.com/psaab/xpf --skill review-pr-psaab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: review-pr
Source: https://github.com/psaab/xpf/tree/main/.claude/skills/review-pr
Command: npx skills add https://github.com/psaab/xpf --skill review-pr-psaab

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill addresses the limitations of single-reviewer PR processes by orchestrating a hostile, multi-perspective review that catches mechanical, numerical, and edge-case bugs often missed by standard human or single-agent reviews.

Core Features & Use Cases

  • Quad-Reviewer Methodology: Coordinates Claude, Codex, Antigravity, and Copilot to perform concurrent, adversarial code analysis.
  • Hostile Verification: Forces reviewers to prove claims against the actual source code, preventing hallucinations and stale feedback.
  • Use Case: Use this skill when reviewing complex network-dataplane changes in the xpf codebase to ensure concurrency, numerical stability, and BPF verifier compliance before merging.

Quick Start

Invoke the review-pr skill with the specific GitHub pull request number to initiate the automated hostile review cycle.

Frequently Asked Questions about review-pr

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

FAQPage Schema
What is multi-agent adversarial code review for GitHub pull requests?▼

Adversarial code review orchestrates multiple specialized AI agents to perform concurrent, hostile verification of GitHub pull requests, forcing cross-referenced checks against actual source code to catch mechanical, numerical, and concurrency bugs.

How do I automate a hostile PR review for network-dataplane changes?▼

You automate a hostile PR review by invoking the skill with a specific GitHub pull request number, which triggers a quad-reviewer methodology across Claude, Codex, Antigravity, and Copilot to analyze complex network-dataplane changes.

Does multi-agent code review work for Rust and Go system-level changes?▼

Multi-agent code review specifically supports high-integrity validation of complex system-level changes written in Rust and Go, analyzing network-expert codebases for concurrency vulnerabilities and numerical edge-case bugs.

What is the best way to prevent hallucinations during automated code review?▼

Preventing hallucinations during automated code review requires hostile verification, a mechanism that forces AI reviewers to prove their claims against the actual source code, ensuring feedback is accurate and not stale.

When do I need an adversarial multi-agent review instead of a single-agent review?▼

An adversarial multi-agent review is needed when standard single-reviewer processes fall short, specifically to catch mechanical, numerical, and edge-case bugs in complex network-dataplane changes before merging.

Can I use this to verify BPF verifier compliance in the xpf codebase?▼

Yes, you can use this skill to review complex network-dataplane changes in the xpf codebase, specifically ensuring concurrency, numerical stability, and BPF verifier compliance are met before merging.

Related Skills

Explore More Software Engineering Skills→
dyoshikawadyoshikawa
1

explain-pr

Analyze GitHub pull request metadata and diffs to summarize problem backgrounds and technical solutions.

Community
Basic
zach-sourcezach-source

ci-polling

Monitor GitHub Actions and GitLab CI/CD pipeline statuses with background polling tasks.

Community
Intermediate
zajcazajca
2

go-linting

Configure golangci-lint with recommended Go linters for static analysis.

Community
Basic
fcouryfcoury
1

brutal-review

Review code changes across logic, architecture, and testing using VCS commands.

Community
Advanced