
Structural code security for MCP: block dangerous code before execution.
Visit Code Firewall McpCode Firewall Mcp is a security filter for Model Context Protocol (MCP) that detects and blocks potentially dangerous code patterns before they reach execution tools. It uses tree-sitter for parsing, normalizes code to abstract away specifics, and embeds the structure for comparison against a blacklist of unsafe patterns using embeddings provided by Ollama and vector storage with ChromaDB. Designed for developers running code execution agents (especially in environments like Claude Code or Claude Desktop), it aims to proactively prevent arbitrary command execution and related threats.
Visit Code Firewall Mcp's official website for product details and getting started.
Join the community to discuss issues, share solutions, and collaborate on improvements.
Comprehensive documentation on how to implement and use Code Firewall Mcp effectively.
Explore practical examples and code snippets for integrating Code Firewall Mcp in your projects.