Cursor

Connect Crow to Cursor, the AI-powered code editor.

Option A: Local (stdio)

Run Crow servers locally for direct integration.

Setup Steps

1. Clone and set up Crow:

   git clone https://github.com/kh0pper/crow.git
   cd crow
   npm run setup

2. Create .cursor/mcp.json in your project root:

   {
     "mcpServers": {
       "crow-memory": {
         "command": "node",
         "args": ["/path/to/crow/servers/memory/index.js"],
         "env": {
           "CROW_DB_PATH": "/path/to/crow/data/crow.db"
         }
       },
       "crow-projects": {
         "command": "node",
         "args": ["/path/to/crow/servers/research/index.js"],
         "env": {
           "CROW_DB_PATH": "/path/to/crow/data/crow.db"
         }
       }
     }
   }

3. Restart Cursor to load the MCP servers.

Option B: Remote (HTTP)

Connect to a deployed Crow gateway.

Setup Steps

1. Deploy Crow (Cloud Deploy Guide)

2. Create .cursor/mcp.json:

   {
     "mcpServers": {
       "crow-memory": {
         "url": "https://your-crow-server/memory/mcp"
       },
       "crow-projects": {
         "url": "https://your-crow-server/projects/mcp"
       },
       "crow-tools": {
         "url": "https://your-crow-server/tools/mcp"
       }
     }
   }

3. Cursor will handle the OAuth flow automatically.

Transport

• Local: stdio
• Remote: Streamable HTTP with OAuth 2.1

Cross-Platform Context

Crow automatically delivers behavioral context when Cursor connects — memory protocols and session management are active from the first message.

IDE platforms get minimal transparency output — only Tier 2 checkpoints. MCP prompts (session-start, crow-guide, etc.) are available for deeper guidance. Memories and projects stored from Cursor are shared with all other connected platforms. See the Cross-Platform Guide.

Context Optimization

Cursor uses stdio transport locally. For a lighter setup, crow-core provides a single combined entry point that activates servers on demand instead of running all of them simultaneously. Generate a combined config with:

npm run mcp-config -- --combined

This creates a single crow-core entry in .mcp.json instead of separate entries for each server. For remote deployments, the /router/mcp endpoint offers similar consolidation. See the Context & Performance guide for details.

Verification

In Cursor's AI chat, try:

"Store a memory that Cursor is connected to Crow."