MCP Integration

Commons exposes a full Model Context Protocol server. Connect any MCP-compatible AI tool — Claude, Cursor, Windsurf, Copilot, or your own.

Streamable HTTP26 tools available

What your AI tool can do

Read teams, projects & updates

Query org structure, project statuses, and recent activity.

Browse standups & retros

Get submissions, insights, action items, and team health.

Post updates & comments

Create updates and reply to threads without leaving your tool.

Generate reports

Built-in prompt templates for weekly rollups, project briefings, and health checks.

Setup

  1. 1

    Create an API key

    Go to your organization settings → API Keys and create a key with the "MCP Server" permission preset.

  2. 2

    Add the server to your MCP config

    Most MCP clients use a JSON config file. Add Commons as a server:

    {
  "mcpServers": {
    "commons": {
      "url": "https://trycommons.com/api/mcp/http",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

    Endpoint: https://trycommons.com/api/mcp/http

  3. 3

    Start using it

    Your AI tool will discover Commons's tools, resources, and prompt templates automatically.

Compatible tools

Any tool that supports the MCP protocol works with Commons. Some popular ones:

Claude DesktopAnthropic's desktop app
Claude CodeCLI for software development
CursorAI-powered code editor
WindsurfCodeium's AI IDE
VS Code + CopilotGitHub Copilot with MCP support
Your own clientAny MCP SDK implementation

Security

  • •API key auth — All requests require a valid key. No OAuth flow needed.
  • •Scoped permissions — Each key only accesses what you allow.
  • •Revocable anytime — Rotate or delete keys from org settings.
  • •HTTPS only — All traffic is TLS-encrypted.

For Claude-specific setup (Desktop and Code), see the Claude integration page.