review

Review code with multi-agent analysis and git worktrees.

12|2|Updated Jan 27, 2026
One-click install
npx skills add https://github.com/jikig-ai/soleur --skill review-jikig-ai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/jikig-ai/soleur/tree/main/plugins/soleur/skills/review
Command: npx skills add https://github.com/jikig-ai/soleur --skill review-jikig-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires semgrep, brew, pipx, python3, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill performs deep, deterministic code and document reviews using parallel specialist agents and isolated git worktrees to surface security, performance, architectural, and quality issues that standard lightweight reviews miss.

Core Features & Use Cases

  • Multi-agent analysis: Spawns targeted reviewers (security, performance, architecture, test design, etc.) and synthesizes their findings into prioritized actions.
  • Worktree-based local inspection: Creates isolated git worktrees or checks out PR branches to run local linters, tests, and deterministic SAST (semgrep) without disrupting the main branch.
  • Governed remediation workflow: Applies inline fixes where appropriate, creates well-structured GitHub issues for scope-outs, and enforces provenance and gating rules for merge readiness.

Quick Start

Invoke the review skill with a PR number or GitHub URL to run a full multi-agent analysis, create inline fixes and code-review issues, and produce a prioritized findings summary.

Frequently Asked Questions about review

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

FAQPage Schema
How do I run a multi-agent code review on a GitHub pull request?

Run a multi-agent code review by invoking the skill with a PR number or GitHub URL. It spawns targeted specialist agents, creates an isolated git worktree, runs semgrep SAST, applies inline fixes, and produces a prioritized findings summary.

What is exhaustive multi-agent code review and how does it detect security issues?

Exhaustive multi-agent code review spawns parallel specialist agents to analyze code for security, performance, and architecture issues. It uses deterministic SAST via semgrep and git worktrees to surface problems that standard lightweight reviews miss.

Do I need semgrep and gh CLI authenticated to perform local SAST code reviews?

Yes, you need a git repository with gh CLI authenticated and permission to create worktrees. Semgrep is optional but recommended for deterministic SAST, and you can install it via the provided ensure-semgrep.sh script.

Does this multi-agent code review work with git worktrees without disrupting the main branch?

Yes, the code review uses worktree-based local inspection to check out PR branches and run local linters, tests, and semgrep SAST in isolation. This ensures your main branch remains undisrupted during the analysis.

Can I use this tool to create GitHub issues for scope-outs found during PR code review?

Yes, the governed remediation workflow applies inline fixes where appropriate and automatically creates well-structured GitHub issues for scope-outs. It enforces provenance and gating rules to determine merge readiness.

What is the best way to perform deep code analysis on markdown documents and branches?

Use multi-agent analysis with git worktrees to perform deep local inspection on markdown documents and branch inspections. This synthesizes findings from targeted reviewers into prioritized actions for comprehensive quality checks.