TOOLS · AI-TOOLS · OPEN SOURCE

Review every PR in passes. Brief it in plain words.

AI-Tools is our open-source toolkit of skills for Claude Code and other AI coding agents: multi-pass code review, decision-level PR briefs and Figma to code.
MIT licensed. Runs in your terminal and in GitHub Actions.

/plugin marketplace add enpitech/ai-tools

Install the plugin View it on GitHub ↗

PULL REQUEST · CONVERSATION

/cr-react

Code Review

SEVERITYCOUNTCRITICALWARNING

To fix all issues below, reply with /fix-all

[WARNING]

Issue:

Why:

SUGGESTED FIX

To fix this issue, reply with /fix

/fix

Fix applied and pushed.

A simplified recreation of a cr-react review on a pull request. Placeholder bars stand in for counts, paths and finding text.

WHAT IT IS

A process your agent follows, not a prompt.

AI-Tools is a set of skills for Claude Code and other AI coding agents. Each skill is a repeatable, step-by-step process: what to read, which passes to run, what to report and what to leave out.

The review criteria live in plain Markdown files. Read them, change a threshold, add a pass. The skills run locally in your terminal and in CI as GitHub Actions.

ENGINEERING LEADERS

A review standard, written down. The same passes on every pull request, with criteria in Markdown your team can edit.

REVIEWERS

Start with context, not a wall of diff. A one-screen brief first, then only the findings worth your time.

PMS, QA AND FOUNDERS

Read a PR without reading code. The brief opens with a plain-language summary, with no file or function names.

THE TOOLKIT

Review, brief, build.

Three skills cover the life of a pull request: the review, the summary a whole team can read, and the implementation from a Figma design.

ALSO IN THE BOX

Dependency audit
cr-deps
Six passes over your manifests and lockfiles: vulnerabilities, outdated, deprecated, licenses, unused, lockfile integrity.
Full-codebase audits
cra-*
Every file instead of a diff, plus system checks such as dead exports and circular dependencies, and the dependency audit. Local only.
Auto-detection
cr-fullstack · cr-general
Detects your languages and frameworks, then applies the right criteria to each layer of the stack.
CI-ready
GitHub Actions
One workflow file. Collaborators comment a trigger on a pull request and the review or brief arrives as comments.

CODE REVIEW · CR-*

One pass for each kind of problem.

The cr-* skills review the diff and the files it directly affects in a fixed sequence of passes, and each pass looks for one kind of problem.

React, Node.js and Python get seven passes. Any other language gets five.
Only CRITICAL and WARNING findings at 8/10 confidence or higher. No style nits, no praise.
Each finding names the file and line, says why, and carries a suggested fix.

Agents now write more code than people can read. Anthropic’s write-up on how the Claude Code team does code review makes the case for layered review, where machines do most of the reading and people keep the risky calls.

Every pass, the CI workflow and /fix

CR-REACT · 7 PASSES

1
PASS 1 · BUGS
Null access, race conditions, wrong conditionals
2
PASS 2 · SECURITY
XSS, exposed secrets, unauthenticated Server Actions
3
PASS 3 · COMPONENT ARCHITECTURE
God components, prop drilling, cross-feature imports
4
PASS 4 · HOOKS & STATE MANAGEMENT
Derived state in useEffect, stale closures, missing cleanup
5
PASS 5 · PERFORMANCE
Sequential awaits, missing dynamic imports, barrel imports
6
PASS 6 · CODE QUALITY (React)
Array mutation, missing error boundaries, duplicated logic
7
PASS 7 · INTENT CHECK
Unrelated changes that snuck into the diff
REPORT ONLYCRITICALWARNINGat 8/10 confidence or higher

The seven passes of the React and Next.js review, by their names in the criteria file.

PR COMMENT · CR-BRIEF

/cr-brief

PR Brief

PLAIN-LANGUAGE SUMMARY

In plain words:

REVIEWER SECTION

The decision

Before → After

Before:

After:

Files that matter

FILEWHY

Real risks

Open questions

A simplified recreation of the comment cr-brief posts, with the section names from its template. Placeholder bars stand in for the text.

PR BRIEF · CR-BRIEF

The meat of a PR, on one screen.

A large pull request lands. Someone comments /cr-brief, and one comment arrives: a summary a PM or QA can read, then what was decided, what behaves differently, the files the reviewer must actually open, and the real risks.

It summarizes, it does not review, so it never reports findings. The whole brief stays under 250 words, not counting the files table, and intent nobody documented is called out rather than invented.

The five passes and the hard rules

INSTALL

Three commands in Claude Code.

The toolkit ships as a Claude Code plugin. The same skills install into other agents with the skills CLI.

Claude Code plugin

01 · ADD THE MARKETPLACE
/plugin marketplace add enpitech/ai-tools
02 · INSTALL THE PLUGIN
/plugin install enpitech@enpitech
03 · ACTIVATE THE SKILLS
/reload-plugins

Skills then run as /enpitech:<skill-name>, for example /enpitech:cr-react.

Cursor, Copilot and other agents

Install the skills with the skills CLI. The review criteria are plain Markdown, so any agent that reads files can follow them.

npx skills add enpitech/ai-tools -g -y

GitHub Actions

Copy claude-code-review.yml into .github/workflows/. Plugins do not install workflow files.
Add ANTHROPIC_API_KEY to the repository’s Actions secrets.
Comment /cr-react or /cr-brief on a pull request.

START HERE

Your next review is one comment away.

Free and open source under the MIT license. Read the criteria, change a threshold, send a pull request.

/plugin marketplace add enpitech/ai-tools

View it on GitHub Read the review criteria ↗

Want review standards like these running on your own pipeline? Tell us how your team ships ↓

Contact

Ready to shift?

We accelerate Frontend delivery. The Frontend Delivery Factory installs the system that lets everyone in your org ship production Frontend.

Senior engineers, AI-native by default, shipping production code in your repo. Fast onboarding. Honest work.

What happens next

01
You send the form.
A few lines on your product and timeline.
02
We reply within a day.
Usually faster. No forms after that.
03
We meet and plan.
Scope, team fit, start date. Then engineers.