Can I use Crit for things other than code?

Yes, Crit is excellent for reviewing implementation plans or documentation in markdown format before the agent proceeds with execution.

Can I share the review with other team members?

Yes, the skill includes a share command that generates a public URL and a QR code for sharing the current review state.

How does Crit interact with Claude Code?

Crit starts a local server and opens a browser window for you to review Claude's work. Claude waits for you to click 'Finish Review' before it reads your feedback and starts making changes.

Does Crit support git-based reviews?

Yes, if no specific file is provided, Crit automatically detects uncommitted changes or compares your current branch against the default branch.

What happens after I leave a comment in the Crit UI?

Once you finish the review, Claude reads the comments, applies the requested fixes to your files, and uses the Crit CLI to reply to your comments with status updates.

Crit Review Loop

Name: Crit Review Loop
Author: tomasz-tomczyk

bytomasz-tomczyk

•

155

•

协作工具

Facilitates an interactive, browser-based feedback loop for reviewing AI-generated code changes and implementation plans.

The Crit skill integrates a professional code review workflow directly into the AI agent's lifecycle, allowing for structured human-in-the-loop oversight. When triggered, it launches a browser interface where developers can leave inline comments and feedback on code diffs or markdown plans. Claude then blocks its execution until the review is complete, at which point it automatically parses the feedback, applies the necessary corrections, and replies to each comment. This ensures a high-quality, iterative development process where human guidance and AI execution are perfectly synchronized.

主要功能

01Interactive browser-based UI for inline code and plan reviews

02Automated detection of git changes, branch diffs, or specific plan files

03Live-reloading interface to see applied fixes in real-time

04Structured JSON feedback parsing for precise, multi-file updates

05Automated reply system to confirm changes against specific review comments

06155 GitHub stars

使用场景

01Providing granular feedback on large AI-generated refactors or features

02Collaborative debugging where developers guide the agent via line-specific comments

03Reviewing and refining complex implementation plans before code is written

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tomasz-tomczyk/crit crit

For use in Claude.ai and ChatGPT