Can Crit detect my local git changes?

Yes, if run without arguments, Crit automatically detects uncommitted changes or compares your current branch against the default branch.

How does the AI handle multiple review comments?

Crit outputs a structured JSON file containing all comments. The AI reads this file, addresses each unresolved item, and replies to them via the CLI.

Can I use Crit for implementation plans?

Absolutely. You can run 'crit ' to review a markdown-based implementation plan before the AI starts writing any code.

What is Crit in the context of Claude Code?

Crit is a feedback loop tool that allows you to review an AI agent's work using an interactive web interface rather than just text commands.

How do I signal that I am done reviewing?

Once you have finished leaving inline comments in the browser, click the 'Finish Review' button; the skill will then resume and process your feedback.

Crit Code Review

Name: Crit Code Review
Author: tomasz-tomczyk

bytomasz-tomczyk

•

155

•

コラボレーションツール

Enables an interactive feedback loop for reviewing AI-generated code changes and plans through an inline browser-based interface.

Crit is a specialized collaboration bridge that facilitates real-time human oversight of AI agent activities. By launching a localized web interface, it allows users to leave precise, inline comments on code diffs or implementation plans. This skill ensures that Claude pauses execution until a review is finalized, then automatically processes structured JSON feedback to address suggestions and apply changes iteratively. It is an essential tool for developers who want to maintain high code quality and provide granular guidance to AI agents throughout the development lifecycle.

主な機能

01URL and QR code sharing for collaborative review sessions

02Blocking workflow that prevents AI progress until human feedback is submitted

03Structured JSON feedback processing for precise, quote-based code updates

04Automatic detection of uncommitted changes and branch-level diffs

05155 GitHub stars

06Interactive browser-based interface for inline code and plan comments

ユースケース

01Verifying complex architectural plans before an AI agent begins automated refactoring

02Providing granular feedback on generated diffs to ensure compliance with project standards

03Sharing live review links with team members to gather multi-stakeholder input on AI work

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tomasz-tomczyk/crit crit

For use in Claude.ai and ChatGPT