AI Tools Integration

Integrate Cognitive Modules with various AI development tools.

Supported Tools

Tool	Integration Method	Status
Claude Desktop	MCP Server	✅
Cursor	MCP Server	✅
VS Code	Extension (planned)	🚧
JetBrains	Plugin (planned)	🚧

Claude Desktop

Setup

1. Start MCP server:

cog mcp

2. Configure Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "cognitive": {
      "command": "cog",
      "args": ["mcp"]
    }
  }
}

3. Restart Claude Desktop

Usage

In Claude Desktop:

Use the code-reviewer module to review this code:
[paste code]

Cursor

Setup

1. Start MCP server:

cog mcp --port 3000

2. Add to Cursor settings

Usage

In Cursor chat:

@cognitive run code-reviewer on the selected code

Generic Integration

HTTP API

Start HTTP server:

cog serve --port 8000 --cors

Call from any tool:

curl -X POST http://localhost:8000/api/run/code-reviewer \
  -H "Content-Type: application/json" \
  -d '{"code": "def foo(): pass"}'

WebSocket

const ws = new WebSocket('ws://localhost:8000/ws');

ws.send(JSON.stringify({
  type: 'run',
  module: 'code-reviewer',
  input: { code: '...' }
}));

ws.onmessage = (event) => {
  const result = JSON.parse(event.data);
  console.log(result);
};

Benefits

• Structured Output - Get predictable JSON responses
• Validation - Input/output automatically validated
• Confidence - Know how reliable the result is
• Audit Trail - Every call has rationale