π§° Media Kit¶
Everything you need to write about ContextForge MCP Gateway-assets, ready-to-use copy, badges, images, and quick-start commands.
π€ What is MCP (Model Context Protocol)?¶
MCP is an open-source protocol released by Anthropic in November 2024 that lets AI agents communicate with external tools through a standard JSON-RPC envelope. It's often described as the "USB-C of AI"-a universal connector for language models.
It's widely supported by GitHub Copilot, Microsoft Copilot, AWS Bedrock, Google Cloud AI, IBM watsonx, and 15,000+ servers in the community.
β‘ Why it matters¶
- β Standardized interface contracts via typed JSON Schema
- β Supported across the ecosystem - GitHub/Microsoft Copilot, AWS Bedrock, Google Cloud AI, IBM watsonx, AgentBee, LangChain, CrewAI, and more
- β Strong ecosystem - 15,000+ MCP-compatible servers and multiple clients, with announcements from multiple major vendors
β Current challenges¶
- β Fragmented transports: STDIO, SSE, HTTP - with some methods already deprecated
- β Inconsistent authentication: none, JWT, OAuth
- β Operational overhead: managing endpoints, credentials, retries, and logs for each tool
- β Version mismatch: clients and servers may support different MCP versions
π‘ Why ContextForge MCP Gateway?¶
Problem: Most teams build one-off adapters for each tool or model, leading to maintenance burden and slow development.
ContextForge MCP Gateway solves this by proxying all MCP and REST tool servers through a single HTTPS + JSON-RPC endpoint, with discovery, security, and observability built in.
It lets you create Virtual Servers - remixing tools/prompts/resources from multiple servers, introduce strong Auth - and change protocol versions on the fly. It lets you easily create new MCP Servers without having to write any code - by proxing existing REST services.
And is readily available as open source, published a container image and as a Python module published on PyPi - so you can get started with a single command - and scale all the way up to multi-regional Kubernetes clusters.
Pain Point | How Gateway Solves It |
---|---|
Transport fragmentation (STDIO/SSE/HTTP) | Unifies everything under HTTPS + JSON-RPC |
DIY wrappers & retry logic | Automatic, schema-validated retry handling |
Weak auth layers | Built-in JWT (or OAuth) & rate limiting |
No visibility | Per-call and per-server metrics & logging |
Onboarding difficulties | Built-in admin UI for tools, prompts, and resources |
π Sample Announcements¶
π£ Non-Technical Post
Meet ContextForge MCP Gateway: Simplify AI Tool Connections¶
Building AI agents should be easy-but each tool speaks a different dialect.
ContextForge MCP Gateway is a universal hub: one secure endpoint that discovers your tools and works seamlessly with Copilot, CrewAI, LangChain, and more.
"What should be simple often becomes a debugging nightmare. The ContextForge MCP Gateway solves that." - Mihai Criveti
Try it in 60 seconds:
docker run -d --name mcpgateway \
-p 4444:4444 \
-e JWT_SECRET_KEY=YOUR_KEY \
ghcr.io/ibm/mcp-context-forge:0.6.0
Please β the project on GitHub if you find this useful, it helps us grow!
π οΈ Technical Post
Introducing ContextForge MCP Gateway: The Missing Proxy for AI Agents and Tools¶
ContextForge MCP Gateway normalizes STDIO, SSE, REST, and HTTP MCP servers into one HTTPS + JSON-RPC interface with full MCP support.
It includes schema-validated retries, JWT auth, and a built-in catalog UI.
Docker:
docker run -d --name mcpgateway \
-p 4444:4444 \
-e JWT_SECRET_KEY=YOUR_KEY \
ghcr.io/ibm/mcp-context-forge:0.6.0
PyPI:
Please β the project on GitHub if you find this useful, it helps us grow!
π οΈ Connect Cline VS Code Extension to ContextForge MCP Gateway
A great idea is to create posts, videos or articles on using specific clients or with MCP Gateway. Provide details on how to run and register a number of useful MCP Servers, adding them to the gateway, then using specific clients to connect. For example, Visual Studio Cline, GitHub Copilot, Langchain, etc. Example:
Connect your Cline extension to MCP Gateway¶
ContextForge MCP Gateway offers a unified HTTPS + JSON-RPC endpoint for AI tools, making integration seamless-including with Cline, a VS Code extension that supports MCP.
Start the Gateway (Docker):
docker run -d --name mcpgateway \
-p 4444:4444 \
-e JWT_SECRET_KEY=YOUR_KEY \
ghcr.io/ibm/mcp-context-forge:0.6.0
Or install via PyPI:
β Enjoying this? Leave a star on GitHub!
π What is Cline?¶
Cline is a powerful AI coding assistant for VS Code. It supports MCP, allowing it to discover and use tools provided through MCP Gateway.
π Set up JWT Authentication¶
In your Cline settings, add an MCP server:
{
"name": "MCP Gateway",
"url": "http://localhost:4444",
"auth": {
"type": "bearer",
"token": "<YOUR_JWT_TOKEN>"
}
}
Enable the server in Cline-you should see a green "connected" indicator when authentication succeeds.
π Using MCP Tools in Cline¶
With the connection live, Cline can:
- Automatically list tools exposed by the Gateway
- Use simple prompts to invoke tools, e.g.:
Try it yourself-and don't forget to β the project at ContextForge MCP Gateway!
πΌοΈ Logo & Images¶
Asset | URL |
---|---|
Transparent PNG logo | https://ibm.github.io/mcp-context-forge/logo.png |
Hero demo GIF | https://ibm.github.io/mcp-context-forge/images/mcpgateway.gif |
Architecture overview | SVG |
π£ Social Snippets¶
Tweet / X
Twitter / X
π ContextForge MCP Gateway is now open source! One endpoint to unify & secure AI-tool connections (STDIO, SSE, REST). Give it a spin and drop a β β IBM/mcp-context-forge #mcp #ai #tools
Example
Thrilled to share ContextForge MCP Gateway-an open-source hub that turns fragmented AI-tool integrations into a single secure interface with discovery, observability, and a live catalog UI. Check it out on GitHub and leave us a star β! #mcp #ai #tools
Tip
See Social for example articles and social media posts - and add your own there once published!