Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
tool_name | str | - | The exact name of the tool to be called for this step |
parameters | Dict[str, Any] | {} | The dictionary of parameters to pass to the tool |
description | Optional[str] | None | Optional description of what this step accomplishes |
Functions
No specific functions - Single tool call in a high-level plan.title: AnalysisResult sidebarTitle: AnalysisResult
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
evaluation | str | - | Detailed reasoning and evaluation of the last tool’s result |
next_action | Literal['continue_plan', 'revise_plan', 'final_answer'] | - | Agent directive for the orchestrator: continue, revise, or finalize |
reasoning | Optional[str] | None | Additional reasoning for the chosen next action |
Functions
No specific functions - Structured output of an automated analysis step.title: Thought sidebarTitle: Thought
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
reasoning | str | - | Detailed explanation of understanding and strategy |
plan | List[PlanStep] | - | Step-by-step execution plan of tool calls |
criticism | str | - | Self-critique identifying potential flaws or ambiguities |
action | Literal['execute_plan', 'request_clarification'] | 'execute_plan' | Next action: execute the plan or request clarification |
clarification_needed | Optional[str] | None | Specific clarification needed if action is ‘request_clarification’ |
Functions
No specific functions - Initial structured thinking process for the AI agent.title: ExecutionResult sidebarTitle: ExecutionResult
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
success | bool | - | Whether the plan execution was successful |
final_result | Any | - | The final synthesized result |
execution_history | List[Dict[str, Any]] | [] | History of all tool executions |
total_steps | int | - | Total number of steps executed |
revisions | int | 0 | Number of plan revisions made |
Functions
No specific functions - Result of executing an orchestrated plan.title: plan_and_execute sidebarTitle: plan_and_execute
Functions
plan_and_execute
Master tool for complex tasks. Executes multi-step plans sequentially.
Parameters:
thought(Thought): Structured thought object with reasoning, plan, and criticism
str: Placeholder string - actual execution handled by orchestrator
title: Orchestrator sidebarTitle: Orchestrator
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
agent_instance | Any | - | The agent instance |
task | Task | - | The task to execute |
wrapped_tools | Dict[str, Callable] | - | Dictionary of wrapped tools |
is_reasoning_enabled | bool | - | Whether reasoning is enabled |
original_user_request | str | - | The original user request |
execution_history | str | - | Orchestrator’s execution history |
program_counter | int | 0 | Current step in the plan |
pending_plan | List[PlanStep] | [] | The current plan to execute |
revision_count | int | 0 | Number of plan revisions made |
all_tools | Dict[str, Callable] | - | All available tools except plan_and_execute |
Functions
execute
Main entry point for orchestrator execution.
Parameters:
thought(Thought): The initial thought and plan
Any: The final result of execution
_execute_single_step
Execute a single tool step from the plan.
Parameters:
step(PlanStep): The step to execute
Any: The result of the step execution
_inject_analysis
Inject mandatory analysis step after tool execution.
Returns:
AnalysisResult: The analysis result
_handle_reasoning_step
Handle reasoning injection after tool execution.
Parameters:
tool_name(str): Name of the tool that was executedresult(Any): Result of the tool execution
bool: Whether to continue execution
_request_plan_revision
Request revised plan based on execution history.
_synthesize_final_answer
Synthesize final answer based on execution history.
Returns:
Any: The final synthesized answer