enterprise-mcp-bridge

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

enterprise-mcp-bridge

★ 17 Open GitHub ↗

No description available.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Enterprise MCP Bridge

FastAPI-based wrapper that exposes any Model Context Protocol (MCP) server over plain HTTP/JSON with enterprise-grade capabilities for production use.

[![Deploy to Render](https://render.com/images/deploy-to-render-button.svg)](https://render.com/deploy?repo=https://github.com/inxm-ai/enterprise-mcp-bridge)

Read more about deploying to Render [here first →](#deploy-on-render-weather-mcp-example)

📖 **[Full Documentation](https://inxm-ai.github.io/enterprise-mcp-bridge)**

## Why Enterprise MCP Bridge?

Many open source MCPs are designed for local development or simple demos and fall short for real-world applications. This project addresses those gaps with enterprise production features: multi-user sessions, OAuth integration, centralized management, and scalability.

**[Learn more about the motivation →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/explanation/architecture)**

## Key Features

The Enterprise MCP Bridge automatically extends any MCP server with enterprise-grade capabilities:

- 🔐 **Security & Authentication** - Built-in OAuth2 token exchange, automatic token injection and refresh, group-based data access
- 👥 **Multi-User Sessions** - Concurrent isolated sessions with automatic cleanup and pluggable storage backends
- 🌐 **REST API** - Standard HTTP/JSON endpoints with auto-generated OpenAPI docs
- 🎨 **UI Generation** - AI-powered web application generation from natural language prompts
- 🔄 **Workflow Orchestration** - Multi-agent workflows with human-in-the-loop feedback
- 📡 **SSE Streaming** - Real-time updates for long-running operations

**[See all features →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/home)**

## Quick Start

```bash
git clone https://github.com/inxm-ai/enterprise-mcp-bridge.git
cd enterprise-mcp-bridge
pip install app
uvicorn app.server:app --reload
```

Visit [http://localhost:8000/docs](http://localhost:8000/docs) for the interactive API documentation.

**[Complete getting started guide →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/tutorials/getting-started)**

## Documentation

- 📚 **[Tutorials](https://inxm-ai.github.io/enterprise-mcp-bridge/#/tutorials/getting-started)** - Step-by-step guides to get you started
- 🛠️ **[How-To Guides](https://inxm-ai.github.io/enterprise-mcp-bridge/#/how-to/deploy-production)** - Task-focused guides for specific scenarios
- 📖 **[Reference](https://inxm-ai.github.io/enterprise-mcp-bridge/#/reference/api)** - Complete API and configuration reference
- 💡 **[Explanation](https://inxm-ai.github.io/enterprise-mcp-bridge/#/explanation/architecture)** - Deep dives into architecture and concepts

## Common Use Cases

### Running Local MCP Servers

```bash
# Local MCP: Clone the repo you want to run
git clone https://github.com/modelcontextprotocol/servers.git mcp
docker run -p 8000:8000 -e ENV=dev -v $PWD/mcp/src/fetch:/mcp -it ghcr.io/inxm-ai/enterprise-mcp-bridge:latest python -m mcp_server_fetch

# Custom command: Use environment variable
export MCP_SERVER_COMMAND="npx -y @modelcontextprotocol/server-memory"
uvicorn app.server:app --host 0.0.0.0 --port 8000

# Remote MCP: Connect to a hosted MCP endpoint
export MCP_REMOTE_SERVER="https://mcp.example.com"
uvicorn app.server:app --host 0.0.0.0 --port 8000
```

**[Complete MCP configuration guide →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/how-to/remote-mcp-servers)**

### Production Deployment

```bash
# Docker
docker run -it -p 8000:8000 \
  -e MCP_SERVER_COMMAND="npx -y @modelcontextprotocol/server-memory /data/memory.json" \
  -v $(pwd)/data:/data \
  ghcr.io/inxm-ai/enterprise-mcp-bridge:latest
```

**[See deployment guides →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/how-to/deploy-production)** | **[Docker →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/how-to/docker)** | **[Kubernetes →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/how-to/kubernetes)**

### Deploy on Render (Weather MCP Example)

[![Deploy to Render](https://render.com/images/deploy-to-render-button.svg)](https://render.com/deploy?repo=https://github.com/inxm-ai/enterprise-mcp-bridge)

This deploys `ghcr.io/inxm-ai/enterprise-mcp-bridge:latest` with the weather MCP server into a free Render instance. After deployment, you can test the UI Generation by opening `/app/_generated/start` in your Render app URL, define an id and use a prompt like `Show me the air quality for Berlin, New York and Tokio.`

Environment variables used by this setup:

- `TGI_URL` - LLM API base URL (`https://api.openai.com/v1`)
- `TGI_TOKEN` - API token for the LLM provider (**set this as a secret in Render**)
- `DEFAULT_MODEL` - Default model name, for example `gpt-5.2-codex` - will work best with a code-capable model (codex) for the UI generation features

- `MCP_GIT_CLONE` - MCP repo cloned into the container at startup - the mcp that will be running
- `MCP_SERVER_COMMAND` - Command used to start the MCP server (`python -m mcp_weather_server`) you cloned earlier

- `TGI_CONVERSATION_MODE` - LLM API mode (`responses`)
- `MCP_TEST_ACTOR` - Default actor id for testing/demo flows
- `APP_CONVERSATIONAL_UI_ENABLED` - Enables generated conversational UI features
- `GENERATED_WEB_PATH` - Path where generated UI artifacts are stored
- `ENV` - Runtime mode (`dev`)

⚠️ **Token usage warning:** When the service runs, calls to your configured model provider consume tokens and can incur costs. Keep `TGI_TOKEN` secret and monitor usage limits/billing.

⚠️ **Free Render plan limitations:** The free plan has limited resources so it's slow not only because it uses LLMs, shuts down after a period of inactivity, has no permanent storage (so all your sites will be lost), and may have other restrictions. Also, this page is not authenticated, unless you plan to go big on sharing weather dashboards with everyone you may wanna remove it after the test again.

### OAuth Configuration

The bridge includes built-in OAuth2 token exchange for secure access to downstream resources.

**[OAuth configuration guide →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/how-to/configure-oauth)**

## Examples

The repository includes several complete examples demonstrating different capabilities:

- **minimal-example** - Simple starter with the default memory server
- **memory-group-access** - Group-based data isolation
- **token-exchange-m365** - Microsoft 365 integration with token exchange
- **remote-mcp-github** - GitHub MCP server with token exchange
- **remote-mcp-atlassian** - Jira and Confluence integration

**[Browse all examples →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/reference/examples)**

## Configuration Reference

Key environment variables:

- `MCP_SERVER_COMMAND` - Command to launch the MCP server
- `MCP_REMOTE_SERVER` - URL for remote MCP server
- `MCP_BASE_PATH` - Base path for API routing
- `KEYCLOAK_URL`, `KEYCLOAK_REALM` - OAuth configuration
- And many more...

**[Complete configuration reference →](https://inxm-ai.github.io/enterprise-mcp-bridge/#/reference/configuration)**

## Contributing

Contributions are welcome! Please see our **[contributing guidelines](https://inxm-ai.github.io/enterprise-mcp-bridge/#/contributing)** for details.

## License

GPL-3.0 - See **[LICENSE](https://inxm-ai.github.io/enterprise-mcp-bridge/#/license)** for details.

## Support

- 📖 **[Documentation](https://inxm-ai.github.io/enterprise-mcp-bridge)**
- 💬 **[GitHub Discussions](https://github.com/inxm-ai/enterprise-mcp-bridge/discussions)**
- 🐛 **[Issue Tracker](https://github.com/inxm-ai/enterprise-mcp-bridge/issues)**
- 🔒 **Security**: Report vulnerabilities to security@inxm.ai

enterprise-mcp-bridge

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README