home-lab/research/claude-task-master-ai-integration-status.md

Claude Task Master AI - Nix Package & VS Code MCP Integration

Current Status

✅ Completed

1. Nix Package: Successfully built and packaged Claude Task Master AI
2. Local Installation: Binary available at /home/geir/Home-lab/result/bin/task-master-ai
3. Ollama Integration: Configured to use local Ollama models on grey-area:11434
4. VS Code MCP Setup: Configured for integration with Cursor/VS Code

🔧 Configuration Files

Task Master Configuration

• Location: /home/geir/Home-lab/.taskmaster/config.json
• Models:
  • Main: qwen3:4b (general tasks)
  • Research: deepseek-r1:1.5b (reasoning tasks)
  • Fallback: gemma3:4b-it-qat (backup)
• Provider: openai (using OpenAI-compatible API)
• Base URL: http://grey-area:11434/v1

VS Code MCP Configuration

• Location: /home/geir/Home-lab/.cursor/mcp.json
• Command: Direct path to Nix-built binary
• Environment: OpenAI-compatible mode with local Ollama

🎯 Available MCP Tools

The Task Master MCP server provides these tools for AI assistants:

Project Management

• initialize_project - Set up new Task Master project
• models - Configure AI models and check status
• parse_prd - Generate tasks from PRD documents

Task Operations

• get_tasks - List all tasks with filtering
• get_task - Get specific task details
• next_task - Find next task to work on
• add_task - Create new tasks with AI
• update_task - Update existing task
• set_task_status - Change task status
• remove_task - Delete tasks

Subtask Management

• add_subtask - Add subtasks to existing tasks
• update_subtask - Update subtask information
• remove_subtask - Remove subtasks
• clear_subtasks - Clear all subtasks from tasks

Advanced Features

• expand_task - Break down tasks into subtasks
• expand_all - Auto-expand all pending tasks
• analyze_project_complexity - Complexity analysis
• complexity_report - View complexity reports

Dependencies

• add_dependency - Create task dependencies
• remove_dependency - Remove dependencies
• validate_dependencies - Check for issues
• fix_dependencies - Auto-fix dependency problems

🚀 Usage in VS Code

1. Restart VS Code/Cursor after updating .cursor/mcp.json
2. Access via AI Chat: Use Claude or GitHub Copilot
3. Example Commands:
   • "Initialize a new Task Master project in my current directory"
   • "Create a task for setting up a new home lab service"
   • "Show me the next task I should work on"
   • "Expand task 5 into detailed subtasks"

🔍 Current Issue

The Task Master binary works as an MCP server but appears to hang when making AI API calls to Ollama. This might be due to:

1. Network connectivity between the host and grey-area
2. OpenAI API compatibility formatting differences
3. Authentication handling with the fake API key

Workaround: Use Task Master through the MCP interface in VS Code, where the AI assistant can handle the tool calls without direct API communication.

🛠️ Troubleshooting

Check Ollama Connectivity

curl -X POST http://grey-area:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model": "qwen3:4b", "messages": [{"role": "user", "content": "Hello"}]}'

Verify Task Master Tools

cd /home/geir/Home-lab
echo '{"jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {}}' | \
  OPENAI_API_KEY="fake-key" OPENAI_BASE_URL="http://grey-area:11434/v1" \
  ./result/bin/task-master-ai

Check MCP Server Status

In VS Code, open the Output panel and look for MCP server connection logs.

📋 Next Steps

1. Test MCP Integration: Try using Task Master tools through VS Code AI chat
2. Debug API Connectivity: Investigate why Task Master hangs on API calls
3. Create Sample Project: Initialize a test project to validate functionality
4. Documentation: Create user guides for common workflows

🏗️ Architecture

VS Code/Cursor (AI Chat)
    ↓ MCP Protocol
Task Master AI (Nix Binary)
    ↓ OpenAI-compatible API
Ollama (grey-area:11434)
    ↓ Model Inference
Local Models (qwen3:4b, deepseek-r1:1.5b, gemma3:4b-it-qat)

This setup provides a complete local AI-powered task management system integrated with your development environment while maintaining full privacy and control over your data.