CODAP-mcp-bck

chaddorsey/CODAP-mcp-bck

3.2

If you are the rightful owner of CODAP-mcp-bck and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

The CODAP MCP Integration is a fully operational Model Context Protocol server that allows large language models to interact with the CODAP data analysis tool for data manipulation and visualization.

Tools

  1. create_dataset_with_table

    Create datasets with automatic table display.

  2. create_graph

    Generate visualizations like scatter plots and bar charts.

  3. create_data_context

    Create new data contexts.

  4. create_items

    Add cases to existing datasets.

  5. get_data_contexts

    List all data contexts.

  6. get_components

    List all visualization components.

CODAP MCP Integration

A Model Context Protocol (MCP) server for the CODAP (Common Online Data Analysis Platform) data analysis tool, enabling large language models to interact directly with CODAP for data manipulation and visualization.

🌟 Status: Fully Operational

The system is now fully deployed and operational! LLMs can:

  • ✅ Discover tools via the metadata endpoint
  • ✅ Queue tool requests which are processed asynchronously
  • ✅ Create real datasets in your CODAP session
  • ✅ Generate visualizations (graphs, tables, etc.)
  • ✅ Execute all tools through the web-based API

No local installation required - everything runs through web APIs.

🚀 Quick Start for LLMs

  1. Get session code from your CODAP session (e.g., T5SDQIRW)
  2. Discover available tools: 
    GET https://codap-mcp-cdorsey-concordorgs-projects.vercel.app/api/metadata?code=YOUR_SESSION_CODE
  3. Queue tool requests: 
    POST https://codap-mcp-cdorsey-concordorgs-projects.vercel.app/api/request
{
  "sessionCode": "YOUR_SESSION_CODE",
  "requestId": "unique-id-123",
  "toolName": "create_dataset_with_table",
  "params": { ... }
}

🔧 System Architecture

Web-Based Operation

  • Request Queue: Tools are queued via HTTP POST to /api/request
  • SSE Streaming: Browser worker receives requests via Server-Sent Events from /api/stream
  • Tool Execution: Browser worker executes tools directly against CODAP Plugin API
  • Response Storage: Results stored in KV and accessible via web APIs

Available Tools (9 total)

  • create_dataset_with_table - Create datasets with automatic table display
  • create_graph - Generate visualizations (scatter plots, bar charts, etc.)
  • create_data_context - Create new data contexts
  • create_items - Add cases to existing datasets
  • get_data_contexts - List all data contexts
  • get_components - List all visualization components
  • And more...

📊 Example Usage

Create a Student Performance Dataset

POST /api/request
{
  "sessionCode": "T5SDQIRW",
  "requestId": "create-students-001",
  "toolName": "create_dataset_with_table",
  "params": {
    "name": "StudentPerformance",
    "attributes": [
      {"name": "student_id", "type": "categorical"},
      {"name": "math_score", "type": "numeric"},
      {"name": "reading_score", "type": "numeric"}
    ],
    "data": [
      {"student_id": "S001", "math_score": 85, "reading_score": 78},
      {"student_id": "S002", "math_score": 92, "reading_score": 88}
    ]
  }
}

Create a Scatter Plot Visualization

POST /api/request
{
  "sessionCode": "T5SDQIRW", 
  "requestId": "viz-scatter-001",
  "toolName": "create_graph",
  "params": {
    "dataContext": "StudentPerformance",
    "graphType": "scatterplot",
    "xAttribute": "math_score",
    "yAttribute": "reading_score",
    "title": "Math vs Reading Scores"
  }
}

🌐 API Endpoints

EndpointMethodDescription
/api/metadata?code={sessionCode}GETGet available tools and schemas
/api/requestPOSTQueue tool execution request
/api/responsePOSTStore tool execution response
/api/stream?code={sessionCode}GETSSE stream for browser worker

🔧 For Developers

Local Development

npm install
npm run dev

Testing

npm test                    # Unit tests
npm run test:integration   # Integration tests
npm run test:playwright    # E2E tests

Project Structure

  • api/ - Vercel serverless functions (HTTP API)
  • src/services/browserWorker/ - Browser-based tool execution
  • docs/ - API documentation and examples
  • playwright/ - End-to-end tests

📖 Documentation

🔒 Security & Session Management

  • Session codes are 8-character Base32 format (e.g., T5SDQIRW)
  • Sessions auto-expire after inactivity
  • CORS enabled for cross-origin requests
  • Rate limiting on API endpoints

🎯 LLM Integration Patterns

For LLMs with MCP Support

Use the native MCP protocol with tool calling capabilities.

For LLMs without MCP Support

Use the HTTP API endpoints directly:

  1. Call metadata endpoint for tool discovery
  2. POST to request endpoint for tool execution
  3. Monitor results through web interface

For LLMs without Web Access (like ChatGPT)

Use the action-triggered URL patterns or copy-paste workflows described in the integration guide.

📈 Version Management

The API supports version negotiation:

  • Current API Version: 1.0.0
  • Tool Manifest Version: 1.0.0
  • Use Accept-Version header for version control

🤝 Contributing

See project documentation in docs/delivery/ for development workflow and task management.

📄 License

MIT License - see LICENSE file for details.

Ready to integrate with any LLM! 🚀