elang2/mcp-audit-gateway

Overview

A TypeScript gateway for MCP-based AI agents that logs every tool call into an append-only audit trail. Each record is cryptographically signed and hash-chained, making any post-hoc alteration detectable and providing verifiable proof of agent actions.

Best for

Best for
Teams operating MCP agents that need verifiable audit trails for compliance or security

Use cases

• Prove what an agent did in a regulated environment
• Detect tampering with tool-call history
• Audit agent behavior for security investigations

How to use

Install

npx @mcp-audit-gateway/core wrap -- <your mcp server command>

Tested with

Claude Desktop, Claude Code, Cursor

Example client config

{\n  "mcpServers": {\n    "github": {\n      "command": "npx",\n      "args": ["mcp-audit", "wrap", "--", "npx", "@modelcontextprotocol/server-github"]\n    },\n    "filesystem": {\n      "command": "npx",\n      "args": ["mcp-audit", "wrap", "--", "npx", "@modelcontextprotocol/server-filesystem", "/tmp"]\n    }\n  }\n}

Notes

A TypeScript gateway for MCP-based AI agents that logs every tool call into an append-only audit trail. Each record is cryptographically signed and hash-chained, making any post-hoc alteration detectable and providing verifiable proof of agent actions.

0 stars on GitHub. Last updated 2026-09-16. Licensed MIT.

Use cases

• Prove what an agent did in a regulated environment
• Detect tampering with tool-call history
• Audit agent behavior for security investigations

Pros

• Cryptographic signatures and hash-chaining make logs tamper-evident
• Provides verifiable history without relying on agent self-reporting
• TypeScript implementation fits into existing MCP toolchains

Cons

• Zero GitHub stars, so community adoption and review are unproven
• No documented deployment steps or configuration examples provided
• Adds a gateway layer that agents and tools must route through

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.