MCP Code Review

A day after the CLI launch, we packaged the rulebook we use on real projects into a paid pack:

Team Rules Pack — $49, one-time

• 63 validated rules: Python 21, JS/TS 16, Go 13, Java 13

• GitHub Actions + GitLab CI playbook that blocks merges on critical findings

• 20 LLM prompts for review triage

• Drop-in .mcp-code-review.yaml so the whole team reviews against identical rules

The open-source server stays free, local, and now at 700+ PyPI downloads. The pack is for teams that would rather buy 63 tested rules than write them.

Checkout: https://creem.io/checkout/prod_6Z3S3jGNPsCyRSqNi397ZY/ch_6wLlsvodjjvKq73eBpZCP0

CLI mode shipped 🚀 — run aicraft-code-review with no MCP client:

mcp-code-review review-file app.py git diff | mcp-code-review review-diff mcp-code-review review-code "your code here"

CI-friendly exit codes (0 clean / 1 high-medium / 2 critical) make it a natural GitHub Action hook. Custom rules & team-shared profiles are in too — drop a .mcp-code-review.yaml in your repo root and the whole team gets identical rules.

650+ PyPI downloads and counting. Docs + config examples: https://github.com/GoodJobwilliam/aicraft

Hey IH 👋

Quick update on MCP Code Review — the open-source MCP server that reviews your code 100% locally inside Claude Code, Cursor, and Windsurf. No SaaS, no code uploads.

We just shipped two features that came straight from user feedback:

1. Custom rules — define your own regex rules in a .mcp-code-review.yaml file and the reviewer will flag exactly what your team cares about: internal API names, deprecated libraries, naming conventions.

2. Team-shared profiles — point MCP_CODE_REVIEW_CONFIG at a shared config file and every developer reviews against the same rule set.

Everything still runs locally, checks can be scoped per file, and the server handles snippets, diffs, and full repos.

Try it: pip install aicraft-code-review

Would love feedback from anyone doing AI-assisted code review! 🙌