Cody AI Assistant

Overview

Cody is Sourcegraph's AI coding assistant that understands your entire codebase. With Lemonado's MCP integration, Cody can combine deep code understanding with your business data insights for more intelligent assistance.

Prerequisites

• VS Code with Cody extension installed
• Sourcegraph Cody account (Free or Pro)
• Active Lemonado AI Connector with MCP endpoint
• Node.js installed (for MCP server)

Setup Instructions

Step 1: Install Cody Extension

1. Open VS Code
2. Go to Extensions (Cmd+Shift+X or Ctrl+Shift+X)
3. Search for "Sourcegraph Cody"
4. Click Install
5. Sign in to your Cody account

Step 2: Configure MCP for Cody

Cody uses a configuration file for MCP servers. Create or edit:

• macOS/Linux: ~/.config/cody/mcp_servers.json
• Windows: %USERPROFILE%\.config\cody\mcp_servers.json

Step 3: Add Lemonado MCP Configuration

Add this configuration to your mcp_servers.json:

{
  "cody.mcpServers": {
    "lemonado": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-fetch",
        "https://mcp.lemonado.io/mcp"
      ],
      "env": {
        "MCP_AUTH_TYPE": "oauth",
        "MCP_OAUTH_PROVIDER": "lemonado"
      }
    }
  }
}

Step 4: Authenticate with OAuth

When you first use the Lemonado connector, you'll be prompted to authenticate via OAuth. This will open a browser window where you can log in to your Lemonado account and select which AI Connector to use.

Step 5: Restart VS Code

1. Close VS Code completely
2. Reopen VS Code
3. Open Cody chat panel
4. Verify the connection with: @lemonado test connection

Using Lemonado with Cody

Context-Aware Queries

Cody excels at combining code context with data:

How does this CustomerService class relate to @lemonado customer data structure?

Generate a migration to sync this model with @lemonado production schema

Code Generation

Generate code based on real data:

@lemonado Create TypeScript types for our order processing system

Build an API endpoint that matches @lemonado inventory data structure

Data Analysis in Code

Write a function to calculate customer LTV using @lemonado transaction data

@lemonado Generate a report class for our monthly sales metrics

Advanced Features

Cody Commands with MCP

Create custom Cody commands that leverage Lemonado data:

{
  "cody.customCommands": [
    {
      "name": "generate-from-schema",
      "prompt": "Generate code based on @lemonado ${input:table} table schema",
      "context": {
        "selection": true,
        "currentFile": true
      }
    },
    {
      "name": "validate-against-data",
      "prompt": "Validate this code against @lemonado production data patterns"
    }
  ]
}

Multi-Repository Context

Cody can understand relationships across repositories:

How does the API in repo-a interact with @lemonado data used by repo-b?

Enterprise Features

For Cody Enterprise users:

{
  "servers": {
    "lemonado-prod": {
      "command": "node",
      "args": ["..."],
      "permissions": {
        "read": true,
        "write": false
      },
      "audit": true,
      "compliance": {
        "gdpr": true,
        "soc2": true
      }
    }
  }
}

Configuration Options

Environment-Specific Setup

Configure different environments, each with OAuth authentication:

{
  "cody.mcpServers": {
    "lemonado-dev": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-fetch",
        "https://mcp.lemonado.io/mcp"
      ],
      "env": {
        "MCP_AUTH_TYPE": "oauth",
        "MCP_OAUTH_PROVIDER": "lemonado"
      }
    },
    "lemonado-prod": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-fetch",
        "https://mcp.lemonado.io/mcp"
      ],
      "env": {
        "MCP_AUTH_TYPE": "oauth",
        "MCP_OAUTH_PROVIDER": "lemonado"
      }
    }
  }
}

Cody-Specific Features

Smart Context Selection

Cody automatically includes relevant context:

// Cody understands this relates to customers
class CustomerAnalytics {
  // Ask: "@lemonado what fields should this include?"
}

Code Lenses

See data insights directly in your code:

// Cody Lens: 1,234 customers match this query
const premiumCustomers = await db.query(
  'SELECT * FROM customers WHERE tier = "premium"'
);

Inline Completions

Get data-aware completions:

const customer = {
  id: // Cody suggests: UUID format from @lemonado
  name: // Cody suggests: max 255 chars per @lemonado schema
  created: // Cody suggests: ISO timestamp format
};

Best Practices

1. Leverage Cody's Context Engine

Let Cody understand your full application:

Explain how @lemonado order data flows through our microservices

2. Use Cody Commands

Create reusable commands for common tasks:

/generate-dto @lemonado.customers
/validate-schema @lemonado.orders
/create-test-data @lemonado.products

3. Combine with Code Search

Find all files that interact with @lemonado customer data

4. Security Best Practices

Troubleshooting

Cody Not Finding MCP Server

1. Check configuration file location
2. Verify JSON syntax
3. Ensure Node.js is installed
4. Check Cody output panel for errors

Authentication Failures

• Clear browser cookies and re-authenticate
• Check if your AI Connector is active
• Ensure you selected the correct connector during OAuth

Performance Issues

• Enable caching in configuration
• Use specific queries instead of broad searches
• Check VS Code performance monitor
• Upgrade to Cody Pro for better performance

Context Limitations

If Cody's context window is full:

• Use more specific queries
• Clear conversation and start fresh
• Upgrade to Cody Pro for larger context

Integration Patterns

Repository-Aware Queries

How does @lemonado customer data map to our Customer entity across all repos?

Cross-File Understanding

Update all DTOs to match @lemonado latest schema changes

Test Generation

Generate integration tests using @lemonado sample data for edge cases

Documentation

Document this API endpoint with @lemonado real response examples

Cody vs Other AI Assistants

Enterprise Deployment

Self-Hosted Setup

For enterprise customers with self-hosted Sourcegraph:

{
  "cody.mcpServers": {
    "lemonado": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-fetch",
        "https://mcp.lemonado.io/mcp"
      ],
      "env": {
        "MCP_AUTH_TYPE": "oauth",
        "MCP_OAUTH_PROVIDER": "lemonado"
      },
      "proxy": "https://your-sourcegraph-instance.com/proxy"
    }
  }
}

Compliance Features

• SOC 2 Type II compliant
• GDPR data handling
• Audit trail for all queries
• Role-based access control

Team Sharing

Share MCP configurations across your team:

# Export configuration
cody config export --mcp > team-mcp-config.json

# Import on team member's machine
cody config import --mcp team-mcp-config.json

Tips & Tricks

Quick Data Checks

Use Cody's inline chat for quick queries:

// Cmd+I: "@lemonado count of active users"

Refactoring with Data

Refactor this service to handle @lemonado new payment types

Performance Optimization

Analyze this query against @lemonado indexes and suggest optimizations

Schema Validation

Validate our TypeScript interfaces against @lemonado current schema

Next Steps

• Explore Cody documentation
• Learn about MCP protocol