Bernerspace

One MCP server · All your integrations · Zero hassle

<p align="center"> <img src="assets/banner.jpg" alt="Bernerspace Banner" width="800"> </p>

<p align="center"> 🚀 <b>Unified MCP server for all OAuth-enabled integrations</b><br> <i>One server · One JWT · Many tools</i> </p>

<p align="center"> ✅ <b>Current Live Integration:</b> Slack — <b>47 tools</b> tested & available <br> 🔜 <b>Coming Soon:</b> Gmail · Google Calendar · Notion <br> 🎯 <b>Goal:</b> 100+ integrations </p>

Why

Using or integrating MCP servers today is painful because:

• Most MCP servers lack OAuth capabilities, limiting user experience.
• Every MCP server requires separate deployment and management, increasing operational overhead.
• Integration chaos: different authentication layers across MCP servers make unified integration nearly impossible.

What You Get

• Single JWT across services via middleware (HS256, issuer/audience validated).
• OAuth middleware per integration (Slack live; more coming) with DB‑backed token storage.
• Consistent MCP tools interface for each service.
• Unified endpoints you can self‑host, e.g.:
  • http://localhost:8000/mcp/slack
  • http://localhost:8000/mcp/gmail (coming soon)

Current Status

• Current live integration: Slack with 47 fully tested tools available.
  → Learn more about Slack tools: docs/slack/tools.md
• Coming soon: Gmail, Google Calendar, and Notion.

📖 Documentation

All setup instructions, integration guides, and examples are available on our Notion docs:

🔗 Bernerspace Quick Start Docs

The docs are updated regularly with:

• Quick start guides for new integrations
• Step-by-step installation instructions
• Usage examples for each tool
• Roadmap and contribution guidelines

Quickstart

1. Environment

Create a .env with the following variables:

JWT_SECRET=your-jwt-signing-secret
DATABASE_URL=postgresql://localhost:5432/mcp_server
CLIENT_ID=<slack_client_id>
CLIENT_SECRET=<slack_client_secret>
SLACK_REDIRECT_URI=http://localhost:8000/mcp/slack/oauth/callback

1. Install dependencies (choose one)

• Using pip
  • python3 -m venv .venv && source .venv/bin/activate
  • pip install -e .
• Using uv
  • uv venv && source .venv/bin/activate
  • uv pip install -e .

1. Run the server

• Using Python
  • python server.py
  • Server will listen on http://localhost:8000
• Using Docker
  • docker build -t bernerspace-mcp .
  • docker run --env-file .env -p 8000:8000 bernerspace-mcp
  • Server will listen on http://localhost:8000
• Using Docker Compose
  • docker compose up -d

1. Create a JWT to call the MCP server

• python generate_jwt.py --user-id <your_user_id>
• Use the printed token as: Authorization: Bearer TOKEN

OAuth Flow (Slack)

• GET /mcp/slack returns oauth_url and instructions to authorize the workspace.
• Slack redirects to SLACK_REDIRECT_URI (defaults to /mcp/slack/oauth/callback).
• The server exchanges the code, enriches the token details, and persists it in Postgres.
• Tokens are stored in table oauth_tokens with composite key (client_id, integration_type) where client_id = your JWT subject (sub).

Database

• Schema managed with Alembic (migrations included).
• Table: oauth_tokens(client_id, integration_type, token_json, stored_at).
• Configure Postgres via DATABASE_URL.

MCP Client Configuration

Example client entry (mcp.json):

{
  "servers": {
    "slack": {
      "url": "http://localhost:8000/mcp/slack",
      "type": "http",
      "headers": {
        "Authorization": "Bearer YOUR_JWT"
      }
    }
  }
}

If the user hasn’t completed OAuth, tool calls will return an object with requires_auth: true and an oauth_url you can open to complete authorization.

VS Code MCP Client Setup

Use this mcp.json in your VS Code user settings (replace JWT with your generated token):

{
  "servers": {
    "slack": {
      "url": "http://localhost:8000/mcp/slack",
      "type": "http",
      "headers": {
        "Authorization": "Bearer JWT"
      }
    }
  },
  "inputs": []
}

LangChain Example

from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent

client = MultiServerMCPClient(
    {
        "slack": {
            "transport": "streamable_http",
            "url": "http://localhost:8000/mcp/slack",
            "headers": {
                "Authorization": "Bearer YOUR_TOKEN"
            },
        }
    }
)

tools = await client.get_tools()
agent = create_react_agent("openai:gpt-4.1", tools)
response = await agent.ainvoke({"messages": "Can you send hello message to my #general groups?"})

Docker

You can run Bernerspace MCP in Docker:

docker build -t bernerspace-mcp .
docker run --env-file .env -p 8000:8000 bernerspace-mcp

🚀 Roadmap

We’re just getting started. Our goal is to support 100+ integrations so you can connect any third-party service to your MCP server with ease.

Coming Soon

• 📧 Gmail
• 📅 Google Calendar
• 📂 Google Drive
• 🗂️ Notion
• 💻 GitHub
• 💬 Discord
• 📊 Airtable
• 🔍 ElasticSearch
• 📝 Confluence

…and dozens more on the way.

🤝 Contributing

We welcome contributions of all kinds — whether it’s fixing a bug, suggesting a feature, improving documentation, or just sharing feedback.

Ways to Contribute

• ⭐ Star the repo to show your support and help others discover Bernerspace.
• 🐛 Report a bug so we can fix it quickly.
• 💡 Request a feature or suggest an improvement.
• 🛠️ Submit a Pull Request with code changes, docs updates, or tests.

Stay in Touch

• 📧 Email me at ranga.sumit1999@gmail.com
• 🐦 Follow & DM on Twitter (X): @SumitSandbox

Whether you’re a first-time contributor or an experienced developer, we’d love to hear from you and collaborate!

License

This project is licensed under the MIT License.