agentic-community/mcp-gateway-registry

Enterprise-ready MCP Gateway & Registry that centralizes AI development tools with secure OAuth authentication, dynamic tool discovery, and unified access for both autonomous AI agents and AI coding assistants. Transform scattered MCP server chaos into governed, auditable tool access with Keycloak/Entra integration.

What it solves

Many organizations struggle with "credential sprawl" and a lack of centralized governance when deploying AI assets. Instead of every developer manually wiring MCP servers and agents with separate credentials in dotfiles, this project provides a single, governed control plane. It eliminates the need for users to store long-lived tokens on their laptops and provides a unified way to discover and audit AI tools across an enterprise.

How it works

The system consists of a data plane (an nginx reverse proxy for routing, TLS, and authentication validation) and a control plane (a FastAPI service that manages the inventory, access models, and audit trails). It integrates with existing identity providers (like Okta, Keycloak, or Entra ID) via OAuth2/OIDC.

Key mechanisms include:

  • Centralized Gateway: A single entry point for MCP servers, agents, and REST endpoints.
  • Egress Authentication: The gateway brokers third-party SaaS credentials (OAuth 3LO/OBO), injecting tokens on egress so they never reach the user's local machine.
  • Dynamic Discovery: Agents and assistants use natural-language semantic search to find tools at runtime rather than relying on hard-coded configurations.
  • Virtual MCP Servers: The ability to aggregate tools from multiple backends into a single endpoint with per-tool access control.

Who it’s for

  • Developers building AI agents and coding assistants who need a secure way to connect to tools.
  • Platform and Security Teams who need to enforce access control, audit logs, and security scanning for AI assets.
  • Enterprise Decision-Makers looking to standardize the deployment of AI tools across their organization.

Highlights

  • Unified AI Asset Registry: Manages MCP servers, A2A agents, reusable skills, and custom entities in one place.
  • Enterprise-Grade Auth: Supports per-user egress authentication and integration with major identity providers.
  • AI Gateway for HTTP: Can act as an authenticated proxy for model inference (OpenAI, Bedrock) and any generic REST endpoint.
  • Security First: Includes a fail-closed admission gate with security scanning for all registered assets.
  • ** federation**: Can pull in external registries like Anthropic's MCP Registry or AWS Agent Registry.

Related

  • Project
    agentregistry-dev/agentregistry

    An open-source registry for MCP servers, AI agents, skills, and prompts that centralizes discovery, management, and deployment of AI building blocks.

  • Project
    ascending-llc/jarvis-registry

    An enterprise-grade MCP and A2A gateway that provides secure, governed access for AI copilots and autonomous agents to internal enterprise tools and data.

  • Project
    samanhappy/mcphub

    MCPHub is a self-hosted gateway and control plane that unifies, secures, and monitors multiple Model Context Protocol (MCP) servers for AI applications.

  • Project
    ThinkWatchProject/ThinkWatch

    ThinkWatch is a self‑hosted gateway that centralises LLM API calls and AI‑tool (MCP) integrations for organisations. It authenticates via OIDC, enforces per‑user limits, applies security‑guard filters, logs every request to ClickHouse and provides a web console for configuration, cost‑tracking and audit. Deployable with Docker‑Compose or Helm, it supports OpenAI, Anthropic, Gemini, Azure, Bedrock and many third‑party tools.