Skip to content

VS Code + MCP Integration

Transform VS Code into an AI-powered development environment with access to your Nancy Brain knowledge base through the Model Context Protocol (MCP).

What You'll Get

  • Smart Code Assistance: Ask questions about your codebase and get answers from your knowledge base
  • Documentation at Your Fingertips: Instant access to API docs, tutorials, and examples
  • Context-Aware Suggestions: AI suggestions based on your specific repositories and documentation
  • Seamless Workflow: No context switching - everything happens in VS Code

Prerequisites

  • VS Code 1.85.0 or later
  • Nancy Brain installed and knowledge base built
  • MCP-compatible extension (instructions below)

VSCode Native Integraion with Chat (VSCode version>=1.103.0)

Ctrl/Cmd + Shift + P → "MCP: Add Server..."

Option A: Local Command

  • Command: /path/to/env/bin/python -m path/to/run_mcp_server.py

  • Name: nancy-brain

  • Where: Global/Workspace

Option B: Pip

  • Package name: nancy-brain

  • Allow

  • Project: path/to/your/project

  • env: /path/to/nancy-brain

  • KMP_DUPLICATE_LIB_OK

  • Name: nancy-brain

  • Where: Global/Workspace

Option C: Remote HTTP Server

??

Ctrl/Cmd + Shift + P → "MCP: Open User Confoguration"

Global Settings

A local command example;

{
  "servers": {
        "nancy-brain": {
            "type": "stdio",
            "command": "/path/to/env/bin/python",
            "args": [
        "-m",
                "/path/to/nancy-brain/run_mcp_server.py"
            ],
            "env": {
                "PYTHONPATH": "/path/to/project",
                "KMP_DUPLICATE_LIB_OK": "TRUE"
            }
      }
    }
}

Workspace:

A pip example:

```/.code-workspace { "folders": [ { "path": "/path/to/project" } ], "settings": { "mcp.servers": { "type": "stdio", "command": "uvx", "cwd": "${input:cwd}", "args": [ "nancy_brain==0.1.3", "/path/to/run_mcp_server.py" ], "env": { "KMP_DUPLICATE_LIB_OK": "TRUE" } } } } 

^ this doesn't actually work with those settings

---

# For Older Versions of VSCode

## Step 1: Install MCP Extension

Install an MCP-compatible VS Code extension. Popular options:

### Option A: Continue (Recommended)
```bash
# Install the Continue extension
code --install-extension continue.continue

Option B: Codeium

# Install Codeium extension  
code --install-extension codeium.codeium

Option C: Gemini Code Assist

code --install-extension gemini-code-assist.gemini-code-assist

For dedicated Gemini Code Assist Instruction see this page.

Step 2: Build Your Knowledge Base

If you haven't already, create and build a knowledge base:

# Create a knowledge base for your project
nancy-brain init my-project-kb
cd my-project-kb

# Add relevant repositories
nancy-brain add-repo https://github.com/your-org/main-project.git
nancy-brain add-repo https://github.com/your-framework/docs.git

# Build the embeddings
nancy-brain build

Step 3: Configure MCP Connection

For Continue Extension

  1. Open Continue Settings: Ctrl/Cmd + Shift + P → "Continue: Open config.json"

  2. Add Nancy Brain MCP Server: 

    {
  "models": [
    {
      "title": "Claude 3.5 Sonnet + Nancy Brain",
      "provider": "anthropic",
      "model": "claude-3-5-sonnet-20241022",
      "apiKey": "your-anthropic-api-key"
    }
  ],
  "mcpServers": {
    "nancy-brain": {
      "command": "nancy-brain",
      "args": ["serve", "--mcp"],
      "cwd": "/path/to/your/knowledge-base"
    }
  },
  "contextProviders": [
    {
      "name": "mcp",
      "params": {
        "serverName": "nancy-brain"
      }
    }
  ]
}

or, for gemini-code-assist

```~/.gemini/settings.json { "mcpServers": { "nancy-brain": { "command": "/path/to/env/bin/python", "args": [ "/path/to/nancy-brain/run_mcp_server.py" ], "env": { "PYTHONPATH": "/path/to/project", "KMP_DUPLICATE_LIB_OK": "TRUE" } } }, "ideMode": true, "hasSeenIdeIntegrationNudge": true, "selectedAuthType": "oauth-personal" } 

3. **Restart VS Code** to load the new configuration

### For Codeium Extension

1. **Open Codeium Settings**: VS Code Settings → Extensions → Codeium

2. **Add MCP Server**: In the MCP Servers section:
   - **Name**: `nancy-brain`
   - **Command**: `nancy-brain serve --mcp`
   - **Working Directory**: `/path/to/your/knowledge-base`

## Step 4: Test the Integration

### Basic Search Test

1. **Open the AI Chat**: `Ctrl/Cmd + Shift + P` → "Continue: Open Chat" (or equivalent)

2. **Test a query**:
Search the knowledge base for "authentication examples" 
You should see results from your indexed repositories.

### Code Context Test

1. **Open a code file** in your project

2. **Ask a contextual question**:
How do I implement error handling similar to what's shown in the knowledge base? 
The AI should respond with examples from your indexed documentation.

## Example Workflows

### Documentation Lookup
User: How do I configure logging in this framework?

AI: Based on your knowledge base, here are the logging configuration options:

[Results from framework documentation]

Would you like me to show you how to implement this in your current file? 

### Code Examples
User: Show me an example of using the API client

AI: Here's an example from your knowledge base:

[Code snippet from indexed repository]

This shows the typical pattern used in your codebase. Would you like me to adapt this for your current context? 

### Architecture Questions
User: What's the recommended way to structure database models?

AI: According to your project documentation:

[Architecture guidelines from knowledge base]

I can help you implement this pattern in your current file. 

## Advanced Configuration

### Custom Tool Access

Enable specific Nancy Brain tools:

```json
{
  "mcpServers": {
    "nancy-brain": {
      "command": "nancy-brain",
      "args": ["serve", "--mcp", "--tools", "search,retrieve,explore"],
      "cwd": "/path/to/your/knowledge-base",
      "env": {
        "NANCY_LOG_LEVEL": "info"
      }
    }
  }
}

Multiple Knowledge Bases

Connect to different knowledge bases:

{
  "mcpServers": {
    "nancy-frontend": {
      "command": "nancy-brain",
      "args": ["serve", "--mcp"],
      "cwd": "/path/to/frontend-kb"
    },
    "nancy-backend": {
      "command": "nancy-brain", 
      "args": ["serve", "--mcp"],
      "cwd": "/path/to/backend-kb"
    }
  }
}

Performance Tuning

Optimize for your workflow:

{
  "mcpServers": {
    "nancy-brain": {
      "command": "nancy-brain",
      "args": [
        "serve", "--mcp",
        "--cache-size", "1000",
        "--timeout", "30"
      ]
    }
  }
}

Troubleshooting

MCP Server Won't Start

Check Nancy Brain installation: 

nancy-brain --version
nancy-brain serve --help

Verify knowledge base: 

cd /path/to/your/knowledge-base
ls knowledge_base/embeddings/

Check VS Code logs: - Open VS Code Developer Tools: Help → Toggle Developer Tools - Look for MCP connection errors in the console

No Search Results

Test search directly: 

cd /path/to/your/knowledge-base
nancy-brain search "test query"

Check knowledge base content: 

nancy-brain explore --max-depth 2

Connection Timeouts

Increase timeout in config: 

{
  "mcpServers": {
    "nancy-brain": {
      "command": "nancy-brain",
      "args": ["serve", "--mcp", "--timeout", "60"]
    }
  }
}

Permission Issues

Check file permissions: 

chmod -R 755 /path/to/your/knowledge-base

Tips for Best Results

1. Organize Your Knowledge Base

  • Use clear repository names
  • Include relevant documentation repositories
  • Keep the knowledge base focused on your current project

2. Use Specific Queries

  • ❌ "How do I code?"
  • ✅ "How do I implement JWT authentication in this API?"

3. Leverage Context

  • Have relevant files open in VS Code
  • The AI can use both your current code and knowledge base

4. Update Regularly

# Refresh your knowledge base periodically
nancy-brain build --force-update

What's Next?

Need Help?