Briefing

mcp‑code‑sanitizer: AI Code Review via Claude MCP

security
by Sandy · Claude Llama

Use mcp‑code‑sanitizer to review AI‑generated code via Claude MCP.

What to do now

Install mcp‑code‑sanitizer and integrate with Claude.

Summary

mcp‑code‑sanitizer is an MCP server that plugs into Claude Desktop or Cursor, providing a strict AI code review powered by Groq’s llama‑3.3‑70b. It offers tools such as analyze_code, compare_code, explain_code, generate_tests, analyze_file, and generate_report. The server runs on FastMCP, caches results, and can be integrated into GitHub Actions to automatically review PRs.

A real example shows a SQL injection vulnerability detected in 2 seconds with a severity score of 23/100, and the tool suggests a safe parameterized query.

The architecture is lightweight: server.py, config.py, groq_client.py, cache.py, prompts.py, and individual tool files. The GitHub Action posts structured comments and fails the check on critical issues.

This tool is ideal for teams that want an AI‑driven security reviewer that explains why code is unsafe and how to fix it.

Key changes

  • Analyze code for bugs and vulnerabilities
  • Compare code versions to detect regressions
  • Explain code step‑by‑step
  • Generate pytest/jest tests
  • Analyze whole files with parallel chunking
  • Build HTML reports

Affects

internal

Customer impact

Analyzing matches…

Ask about this story

Impact on an agency? Which customers? Compare historically Risks of waiting