aumos-trusted-mcp 0.2.0

Security proxy for MCP (Model Context Protocol) connections

trusted-mcp

Security proxy for MCP (Model Context Protocol) connections

Part of the AumOS open-source agent infrastructure portfolio.

---

Features

• InterceptorChain pipeline that applies an ordered sequence of scanners to every tool call request, response, and tool definition — short-circuits on the first BLOCK result
• Scanner ABC for building custom scanners; ships with RegexScanner, AllowlistScanner, PiiScanner, ArgumentScanner, and DescriptionHashScanner out of the box
• Tool description hash verification detects MCP tool-poisoning attacks where a server's tool descriptions change between sessions
• YAML policy config maps scanner slugs to settings, making the security profile fully declarative and version-controlled
• JSON audit logger writes an append-only record of every PASS/WARN/BLOCK decision with scanner name, reason, tool name, and server name
• Native adapters for Claude Desktop, Cursor, and VS Code that inject the proxy transparently into each client's MCP configuration
• Extensible scanner registry via Python entry-points — third-party scanners install and register without modifying the proxy codebase

Quick Start

Install from PyPI:

pip install trusted-mcp

Verify the installation:

trusted-mcp version

Basic usage:

import trusted_mcp

# See examples/01_quickstart.py for a working example

Documentation

• Architecture
• Contributing
• Changelog
• Examples

Enterprise Upgrade

For production deployments requiring SLA-backed support and advanced integrations, contact the maintainers or see the commercial extensions documentation.

Contributing

Contributions are welcome. Please read CONTRIBUTING.md before opening a pull request.

License

Apache 2.0 — see LICENSE for full terms.

---

Part of AumOS — open-source agent infrastructure.