Mcp Gateway and Registry: Enterprise-Grade Tool Governance for AI Agents

MCP Gateway & Registry is an open-source platform that centralizes access to AI development tools using Anthropic's Model Context Protocol. It provides a unified gateway with enterprise-grade authentication, fine-grained access control, and tool governance for both autonomous AI agents and human developers.

The gateway acts as a reverse proxy that routes requests from AI agents and development tools (VS Code, Cursor, Claude Code) to multiple MCP servers, handling authentication, authorization, and observability transparently.

v1.0.5 just shipped with two major capabilities:

1. *Supply Chain Security* - Automated scanning of MCP servers using Cisco AI Defence MCP Scanner. Every server is scanned for vulnerabilities before agents can access it. Uses YARA pattern detection + LLM-powered threat analysis.

2. *Interactive MCP Registry CLI* - Natural language interface for discovering and using MCP tools. Talk to your registry in plain English ("What servers support OAuth?", "Show me all HTTP transport tools"). Built with Ink, supports Claude and Amazon Bedrock models, with real-time token tracking and cost monitoring.

Key features include: - *Anthropic MCP Registry Compatibility* - Full REST API compatibility. Import curated servers from Anthropic's official registry. - *Enterprise Authentication* - Keycloak integration, OAuth 2.0/3.0, fine-grained access control, audit trails - *Real-Time Observability* - Grafana dashboards with SQLite and OpenTelemetry integration for metrics tracking - *Dual AI Model Support* - Works with Anthropic Claude and Amazon Bedrock models - *Multiple Transport Types* - SSE and Streamable HTTP support

Open source (Apache 2.0), production-ready, and actively developed by the Agentic Community.