Dedalus MCP Documentation Server

Instructions

Analyze documentation for specific tasks (foundation for agent handoffs)

Args:
    task: Analysis task (e.g., "find_gaps", "generate_outline", "check_consistency")
    docs: Optional list of specific documents to analyze
    output_format: Output format (summary, detailed, structured)

Returns:
    Analysis results ready for agent handoff

Implementation Reference

• src/main.py:433-487 (handler)

  The complete implementation of the analyze_docs tool, including the @mcp.tool() decorator for registration, type-hinted parameters serving as input schema, docstring, validation logic, document gathering via list_docs(), and structured output for agent handoffs.

  @mcp.tool()
  def analyze_docs(
      task: str, docs: Optional[List[str]] = None, output_format: str = 'summary'
  ) -> Dict[str, Any]:
      """
      Analyze documentation for specific tasks (foundation for agent handoffs)
  
      Args:
          task: Analysis task (e.g., "find_gaps", "generate_outline", "check_consistency")
          docs: Optional list of specific documents to analyze
          output_format: Output format (summary, detailed, structured)
  
      Returns:
          Analysis results ready for agent handoff
      """
      available_tasks = [
          'find_gaps',
          'generate_outline',
          'check_consistency',
          'extract_examples',
          'identify_prerequisites',
          'suggest_improvements',
      ]
  
      if task not in available_tasks:
          return {
              'error': f'Unknown task. Available tasks: {", ".join(available_tasks)}',
              'available_tasks': available_tasks,
          }
  
      # Gather documents to analyze
      if not docs:
          all_docs = list_docs()
          docs = [doc['path'] for doc in all_docs]
  
      # This is where different analysis agents would be invoked
      # Structure the response for easy handoff to specialized agents
      return {
          'task': task,
          'documents_analyzed': len(docs),
          'output_format': output_format,
          'results': {
              'summary': f"Analysis task '{task}' ready for processing",
              'documents': docs,
              'next_steps': [
                  'Connect specialized agent for this task',
                  'Process documents according to task requirements',
                  'Return structured results',
              ],
          },
          'agent_handoff_ready': True,
          'suggested_model': 'gpt-4'
          if task in ['find_gaps', 'check_consistency']
          else 'claude-3-5-sonnet',
      }

• src/main.py:49-49 (registration)

  The analyze_docs tool is listed in the MCP server instructions as an available tool with its description.

  - analyze_docs(task): Analyze documentation for specific tasks

• src/main.py:539-539 (registration)

  The analyze_docs tool is listed among available tools in the test mode print statement.

  'Tools available: list_docs, search_docs, ask_docs, index_docs, analyze_docs'