tools-only 🧰 Media Kit
Everything you need to write about ContextForge MCP Gateway-assets, ready-to-use copy, badges, images, and quick-start commands.
🤔 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.
⚡ Why it matters
- ✅ Standardized interface contracts via typed JSON Schema
- ✅ Supported across the ecosystem - GitHub/Microsoft Copilot, AWS Bedrock, Google Cloud AI, IBM watsonx, AgentBee, LangChain, CrewAI, and more
- ✅ Strong ecosystem - 15,000+ MCP-compatible servers and multiple clients, with announcements from multiple major vendors
❌ Current challenges
- ❌ Fragmented transports: STDIO, SSE, HTTP - with some methods already deprecated
- ❌ Inconsistent authentication: none, JWT, OAuth
- ❌ Operational overhead: managing endpoints, credentials, retries, and logs for each tool
- ❌ Version mismatch: clients and servers may support different MCP versions
💡 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 |
📑 Sample Announcements
???+ "📣 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](https://github.com/IBM/mcp-context-forge)** 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:**
```bash
docker run -d --name mcpgateway \
-p 4444:4444 \
-e JWT_SECRET_KEY=YOUR_KEY \
-e PLATFORM_ADMIN_EMAIL=admin@example.com \
-e PLATFORM_ADMIN_PASSWORD=changeme \
-e PLATFORM_ADMIN_FULL_NAME="Platform Administrator" \
ghcr.io/ibm/mcp-context-forge:1.0.0-RC-1
```
Please ⭐ the project on GitHub if you find this useful, it helps us grow!
???+ "🛠️ Technical Post" ### Introducing ContextForge MCP Gateway: The Missing Proxy for AI Agents and Tools
**[ContextForge MCP Gateway](https://github.com/IBM/mcp-context-forge)** 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:**
```bash
docker run -d --name mcpgateway \
-p 4444:4444 \
-e JWT_SECRET_KEY=YOUR_KEY \
-e PLATFORM_ADMIN_EMAIL=admin@example.com \
-e PLATFORM_ADMIN_PASSWORD=changeme \
-e PLATFORM_ADMIN_FULL_NAME="Platform Administrator" \
ghcr.io/ibm/mcp-context-forge:1.0.0-RC-1
```
**PyPI:**
```bash
pip install mcp-contextforge-gateway
# Option 1: Use the provided .env.example
curl -O https://raw.githubusercontent.com/IBM/mcp-context-forge/main/.env.example
cp .env.example .env
# Edit .env to customize your settings
mcpgateway --host 0.0.0.0 --port 4444
# Option 2: Set environment variables directly
PLATFORM_ADMIN_EMAIL=admin@example.com \
PLATFORM_ADMIN_PASSWORD=changeme \
PLATFORM_ADMIN_FULL_NAME="Platform Administrator" \
mcpgateway --host 0.0.0.0 --port 4444
```
Please ⭐ the project on GitHub if you find this useful, it helps us grow!
???+ "🛠️ 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](https://github.com/IBM/mcp-context-forge)** 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):**
```bash
docker run -d --name mcpgateway \
-p 4444:4444 \
-e JWT_SECRET_KEY=YOUR_KEY \
-e PLATFORM_ADMIN_EMAIL=admin@example.com \
-e PLATFORM_ADMIN_PASSWORD=changeme \
-e PLATFORM_ADMIN_FULL_NAME="Platform Administrator" \
ghcr.io/ibm/mcp-context-forge:1.0.0-RC-1
```
**Or install via PyPI:**
```bash
pip install mcp-contextforge-gateway
# Option 1: Use the provided .env.example
curl -O https://raw.githubusercontent.com/IBM/mcp-context-forge/main/.env.example
cp .env.example .env
# Edit .env to customize your settings
mcpgateway --host 0.0.0.0 --port 4444
# Option 2: Set environment variables directly
PLATFORM_ADMIN_EMAIL=admin@example.com \
PLATFORM_ADMIN_PASSWORD=changeme \
PLATFORM_ADMIN_FULL_NAME="Platform Administrator" \
mcpgateway --host 0.0.0.0 --port 4444
```
⭐ Enjoying this? Leave a star on GitHub!
---
#### 🔍 What is Cline?
[Cline](https://cline.bot/) is a powerful AI coding assistant for VS Code. It supports MCP, allowing it to discover and use tools provided through MCP Gateway.
---
#### 🔐 Set up JWT Authentication
In your Cline settings, add an MCP server:
```json
{
"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.
---
#### 🚀 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.:
```
Run the `list_files` tool with path: "./src"
```
* Display results and JSON output directly within the VS Code interface
Try it yourself-and don't forget to ⭐ the project at [ContextForge MCP Gateway](https://github.com/IBM/mcp-context-forge)!
🖼️ 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 |
📣 Social Snippets
Tweet / X
!!! example "Twitter / X" 🚀 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 ⭐ → https://github.com/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 ⭐!
#mcp #ai #tools
!!! tip Examples Posts See Social for example articles and social media posts - and add your own there once published!