Task Researcher

🚀 Task Research

This project focuses on task research, including configuration, task structure, and different research modes.

🚀 Quick Start

.env Configuration File Example

Refer to the .env.example file. Key settings include:

• LLM_MODEL: The main model for tasks, analysis, and subtask generation (non - STORM).
• ANTHROPIC_API_KEY, GOOGLE_API_KEY, etc.: API keys for the main LLMs.
• STORM_RETRIEVER: Selection for the STORM retriever (e.g., bing, you, tavily, etc.).
• BING_SEARCH_API_KEY, YDC_API_KEY, TAVILY_API_KEY, etc.: API keys for the selected STORM retriever.
• BIG_STORM_MODEL (optional): A large model for internal STORM processing.
• SMALL_STORM_MODEL (optional): A small model for internal STORM processing.
• STORM_SEARCH_TOP_K, STORM_MAX_TOKENS_*: Control the depth of STORM.
• File paths (e.g., TASKS_FILE_PATH, etc.).

Task Structure (tasks/tasks.json)

The task structure follows the Pydantic models defined in task_master_py/models.py, including Task, Subtask, and TasksData.

{
  "meta": {
    "projectName": "Task Master Project",
    "projectVersion": "0.1.2",
    "sourceFiles": ["scripts/plan.md", "..."],
    "generatedAt": "...",
    "totalTasks": 15
  },
  "tasks": [
    {
      "id": 1,
      "title": "Set up the project environment",
      "phase": 1,
      "description": "Initialize the project structure, install dependencies, and configure code inspection tools.",
      "details": "...",
      "status": "pending",
      "dependencies": [],
      "priority": "high",
      "testStrategy": "Verify the environment setup by running basic commands.",
      "subtasks": []
    },
    {
      "id": 2,
      "title": "Implement the core data model",
      "phase": 1,
      "description": "Define Pydantic models for tasks and reports.",
      "details": "...",
      "status": "pending",
      "dependencies": [1], // Depends on task 1
      "priority": "high",
      "testStrategy": "Write unit tests for model validation.",
      "subtasks": [
        {
            "id": 1, // Subtask ID relative to the parent task
            "title": "Define the task model",
            "description": "...",
            "details": "...",
            "status": "pending",
            "dependencies": [], // Dependencies between subtasks (integers)
            "acceptanceCriteria": "The Pydantic model passes validation...",
        },
        {
            "id": 2,
            "title": "Define the report model",
            "description": "...", 
            "details": "...",
            "status": "pending",
            "dependencies": [1], // Depends on subtask 1
            "acceptanceCriteria": "The Pydantic model passes validation...",
        }
      ]
    },
    // ... More tasks
  ]
}

Difference between --research and --research-hint

• expand --research: Triggers a complete STORM - based workflow: AI generates questions -> AI groups questions -> knowledge - storm performs web searches and synthesizes reports -> AI uses the synthesized reports as context to generate subtasks. Requires knowledge - storm and configured retriever API keys.
• analyze-complexity --research-hint: Modifies the prompt of the main LLM to make it analyze using its internal knowledge base more like a research assistant. Does not use knowledge - storm or perform real - time web searches.
• update --research-hint: Similar to analyze-complexity, it prompts the main LLM to use its internal knowledge when processing update prompts.