Skip to main content

​
Parameters

ParameterTypeDefaultDescription
agentAgentRequiredThe agent instance
taskTaskRequiredThe task being executed
stateOptional[State]NoneGraph execution state

​
Functions

​
get_context_prompt

Public getter to retrieve the constructed context prompt. Returns:
  • str: The constructed context prompt string

​
manage_context

Asynchronous context manager for building the task-specific context. Parameters:
  • memory_handler (Optional[MemoryManager]): Memory manager for context injection
Returns:
  • AsyncContextManager: Context manager that handles context building
Usage: 
async with context_manager.manage_context(memory_handler) as manager:
    context_prompt = manager.get_context_prompt()
    # Use the context prompt

​
get_knowledge_base_health_status

Get health status of all KnowledgeBase instances in the context. Returns:
  • Dict[str, Any]: Health status dictionary for each KnowledgeBase

​
get_context_summary

Get a comprehensive summary of the current context configuration. Returns:
  • Dict[str, Any]: Context summary containing:
    • task: Task information (id, description, attachments, etc.)
    • context: Context items (knowledge_bases, tasks, task_output_sources, etc.)
    • agent: Agent configuration
    • state: State availability
Features:
  • Dynamic Context Building: Constructs context from multiple sources
  • Knowledge Base Integration: Handles RAG queries and static knowledge bases
  • Task Context: Processes Task objects and TaskOutputSource items
  • Error Handling: Comprehensive error handling for all context types
  • Health Monitoring: Provides health status for KnowledgeBase instances
  • Context Formatting: Formats RAG results with metadata and scores
  • Memory Integration: Integrates with MemoryManager for context injection
  • State Support: Handles graph execution state for task outputs
⌘I