grepai-mcp-cursor（grepai MCP Cursor 搜索）

GrepAI MCP Integration with Cursor

This skill covers integrating GrepAI with Cursor IDE using the Model Context Protocol (MCP).

When to Use This Skill

• Setting up GrepAI in Cursor
• Enabling semantic search for Cursor AI
• Configuring MCP for Cursor
• Troubleshooting Cursor integration

What is Cursor?

Cursor is an AI-powered IDE that supports MCP for external tools. GrepAI integration gives Cursor's AI:

• Semantic code search beyond simple text matching
• Call graph analysis for understanding dependencies
• Index-based code navigation

Prerequisites

1. GrepAI installed
2. Ollama running (or other embedding provider)
3. Project indexed (grepai watch)
4. Cursor IDE installed

Configuration

Step 1: Create MCP Config File

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

{
  "mcpServers": {
    "grepai": {
      "command": "grepai",
      "args": ["mcp-serve"]
    }
  }
}

Step 2: Restart Cursor

Close and reopen Cursor for the config to take effect.

Step 3: Verify

Ask Cursor's AI:

"Search the codebase for authentication"

Cursor should use the grepai_search tool.

Global Configuration

For GrepAI in all Cursor projects, use global config:

Location

• macOS: ~/.cursor/mcp.json
• Linux: ~/.cursor/mcp.json
• Windows: %APPDATA%\Cursor\mcp.json

Content

{
  "mcpServers": {
    "grepai": {
      "command": "grepai",
      "args": ["mcp-serve"]
    }
  }
}

Per-Project Configuration

For project-specific settings:

{
  "mcpServers": {
    "grepai": {
      "command": "grepai",
      "args": ["mcp-serve"],
      "cwd": "/absolute/path/to/project"
    }
  }
}

Available Tools

Once configured, Cursor has access to:

Usage Examples

Finding Code

Ask Cursor:

"Find code that handles user login"

Cursor uses grepai_search to find semantically related code.

Understanding Dependencies

Ask Cursor:

"What functions call validateToken?"

Cursor uses grepai_trace_callers to show all callers.

Code Navigation

Ask Cursor:

"Show me the call graph for processPayment"

Cursor uses grepai_trace_graph to display dependencies.

Cursor Settings Integration

Enable MCP in Settings

1. Open Cursor Settings (Cmd+, / Ctrl+,)
2. Search for "MCP"
3. Ensure MCP is enabled

Verify MCP Status

1. Open Command Palette (Cmd+Shift+P / Ctrl+Shift+P)
2. Search "MCP"
3. Check connected servers

Windsurf Configuration

Windsurf uses the same MCP format as Cursor:

Location

Create .windsurf/mcp.json:

{
  "mcpServers": {
    "grepai": {
      "command": "grepai",
      "args": ["mcp-serve"]
    }
  }
}

Multiple Projects Setup

Option 1: Separate Configs

Each project has its own .cursor/mcp.json with appropriate cwd.

Option 2: Workspaces

# Create workspace
grepai workspace create dev
grepai workspace add dev /path/to/project1
grepai workspace add dev /path/to/project2

{
  "mcpServers": {
    "grepai": {
      "command": "grepai",
      "args": ["mcp-serve", "--workspace", "dev"]
    }
  }
}

Environment Variables

If GrepAI uses environment variables (like API keys):

{
  "mcpServers": {
    "grepai": {
      "command": "grepai",
      "args": ["mcp-serve"],
      "env": {
        "OPENAI_API_KEY": "sk-..."
      }
    }
  }
}

Better: Set environment variables in your shell profile instead.

Troubleshooting

MCP Not Recognized

❌ Problem: Cursor doesn't see GrepAI tools

✅ Solutions:

1. Check file location: .cursor/mcp.json in project root
2. Verify JSON syntax (no trailing commas)
3. Restart Cursor completely
4. Check grepai is in PATH

Search Returns Nothing

❌ Problem: Empty search results

✅ Solutions:

1. Ensure index exists: grepai status
2. Run grepai watch first
3. Verify working directory

Connection Errors

❌ Problem: MCP connection failed

✅ Solutions:

1. Test manually: grepai mcp-serve
2. Check Ollama: curl http://localhost:11434/api/tags
3. Look at Cursor's developer console for errors

Wrong Results

❌ Problem: Results from wrong project

✅ Solutions:

1. Set explicit cwd in config
2. Check you opened the right folder in Cursor
3. Use grepai_index_status to verify

Performance Tips

1. Background daemon: Keep grepai watch --background running
2. Use compact mode: MCP tools use compact by default
3. Limit results: AI will request appropriate limits
4. Index regularly: Especially after git pull

Comparison: Cursor vs Claude Code

Best Practices

1. Version control: Add .cursor/mcp.json to git (without secrets)
2. Team setup: Document MCP config in README
3. Keep index fresh: Run watch daemon
4. Test locally: Verify grepai mcp-serve works first
5. Use workspaces: For multi-project setups

Removing Integration

Delete .cursor/mcp.json and restart Cursor.

Or remove just GrepAI:

{
  "mcpServers": {
    // Remove grepai entry
  }
}

Output Format

Successful Cursor setup:

✅ GrepAI MCP Integration for Cursor

   Config: .cursor/mcp.json
   Server: grepai mcp-serve
   Status: Ready

   Available tools:
   - grepai_search
   - grepai_trace_callers
   - grepai_trace_callees
   - grepai_trace_graph
   - grepai_index_status

   Cursor AI can now search your code semantically!

   Test: Ask Cursor "search for authentication code"