Show HN: Paste your MCP config, see what each server does at startup

Category: security

Tags: mcp, security-audit, static-analysis, documentation-verification

Score: 5.0/10 (Innovation: 5, Technical: 6, Documentation: 7, Utility: 4)

This project analyzes 67 MCP (Model Context Protocol) servers by starting them in a network-isolated environment and comparing their startup behavior—such as tools offered and websites contacted—against their README documentation. It reveals discrepancies like undocumented tool names and network calls, providing verifiable evidence for each finding. It's interesting for its empirical, safety-focused approach to auditing AI tool integrations.

Target audience: security researchers, AI developers, DevOps engineers

Repository: https://mcp.brydge-ai.com

View on Hacker News