claude-review

Review plans, diffs, and code via the local claude CLI.

5|Updated Apr 15, 2026
One-click install
npx skills add https://github.com/found-cake/claude-review-skill --skill claude-review-found-cake
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: claude-review
Source: https://github.com/found-cake/claude-review-skill/tree/main/skill
Command: npx skills add https://github.com/found-cake/claude-review-skill --skill claude-review-found-cake

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a local Claude-based second opinion for reviewing plans, diffs, and code via the claude CLI, helping teams catch issues early and improve overall quality.

Core Features & Use Cases

  • Plan review: validate implementation strategies before coding.
  • Code/diff review: assess diffs or full codebases for correctness and quality.
  • Exploit-chain validation: verify attack logic or CTF chains with Claude as a sanity check.
  • Non-interactive execution: run Claude non-interactively and summarize results for quick decision-making.

Quick Start

Invoke Claude locally to review plans, diffs, and code using the CLI.

Frequently Asked Questions about claude-review

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

FAQPage Schema
How do I get a Claude second opinion on a git diff for code review?

To get a Claude second opinion on a git diff, invoke the local Claude CLI to review your code changes. The Skill requires git available for diff-based review, parsing stream-json output with jq to assess correctness and quality.

Can I validate an exploit chain using the Claude CLI?

Yes, you can validate an exploit chain using the Claude CLI. The Skill supports exploit-chain verification by running Claude non-interactively to check attack logic or CTF chains, providing a sanity check on security workflows.

What dependencies do I need to review plans and code with Claude locally?

To review plans and code with Claude locally, you need the claude CLI in PATH and jq for stream-json parsing, with git available for diff-based review. These dependencies enable non-interactive execution and result summarization.

How does non-interactive Claude execution work for plan validation?

Non-interactive Claude execution for plan validation runs the local CLI without user interaction to assess implementation strategies. It summarizes results for quick decision-making, allowing teams to catch issues early before coding begins.

What is the best way to review a full codebase with Claude for correctness?

The best way to review a full codebase with Claude is using the local CLI to obtain a second opinion on development work. This Skill assesses codebases or diffs for correctness and quality, helping improve overall project standards.

Why does the Claude review Skill require jq for stream-json parsing?

The Claude review Skill requires jq for stream-json parsing to process the non-interactive output generated by the local Claude CLI. This allows the Skill to extract and summarize review results efficiently for quick decision-making.