MCP Context Forge by Ibm is an open-source AI gateway, registry, and proxy built with Python, FastAPI, and SQLAlchemy. It is available as a PyPI package and OCI images. This platform connects and federates various AI services, enabling a unified interface for diverse tools. It supports multiple transport protocols and integrates with many AI clients, including OpenAI, Anthropic, Claude Desktop, and IBM Bob. The system provides secure access and management capabilities for AI agent interactions, making it a central point for AI tool orchestration.

What does MCP Context Forge connect to?

MCP Context Forge functions as an AI gateway, registry, and proxy. It establishes connections and federates Model Context Protocol (MCP) servers, Agent-to-Agent (A2A) servers, and various REST/gRPC APIs. This capability allows it to unify disparate AI services and tools under a single management point. Its primary role is to provide a centralized access layer for different AI contexts and external service integrations, supporting a diverse ecosystem of AI interactions and communication flows.

Core features and tools

The program acts as a Tools Gateway, managing and exposing tools from federated MCP, A2A, and REST/gRPC services. It performs gRPC-to-MCP translation using automatic reflection-based service discovery, simplifying integration. MCP Context Forge virtualizes legacy APIs, presenting them as MCP-compliant tools and servers, extending their utility to modern AI agents. It does not offer its own named callable tools, but provides a unified interface for tools from integrated services. Supported transport protocols include HTTP, JSON-RPC, WebSocket, SSE, stdio, and streamable-HTTP.

Who is it for?

MCP Context Forge is designed for enterprise teams adopting Model Context Protocol, developers, and experienced DevOps teams requiring full control and Kubernetes capacity. It is suitable for those needing to manage and orchestrate AI tools and agents across various services and platforms. This includes organizations looking to centralize AI client access to diverse tools and virtualize existing APIs for AI interaction. Its capabilities are valuable for complex AI environments. Skip this if your team does not require federated AI service management or Kubernetes-level control.

How to get started

The server authenticates using environment variables for various schemes, including Basic Auth, JWT Bearer tokens, and custom headers, and supports OAuth-scoped authentication with rate limiting. Setup involves configuring environment variables such as BASIC_AUTH_PASSWORD and AUTH_ENCRYPTION_SECRET, and it supports transport protocols like HTTP, alongside multi-database compatibility with PostgreSQL, MySQL, and SQLite.

Use cases

• Federating multiple MCP and REST services to provide a single, unified endpoint for AI clients to access diverse tools and capabilities efficiently.
• Virtualizing legacy APIs as MCP-compliant tools, allowing AI agents to interact with existing REST APIs as native MCP tools, extending their lifespan.
• Optimizing agent and tool calling through centralized discovery, guardrails, management, rate limiting, and observability for AI agents, enhancing performance.

MCP Context Forge offers a centralized solution for managing and connecting diverse AI services and tools. It enables the virtualization of legacy APIs and provides a unified gateway for AI clients. With support for multiple authentication methods and database compatibility, it serves as a foundational component for AI orchestration.