Skip to content

Planner Meta Agent Configuration

Meta Planner Agent acts as the central orchestrator, coordinating multiple specialized agents (like ReAct, Multi-Agent, or hybrids) to solve complex queries. It extends the Meta Agent by using a multi-system-prompt approach for finer control and supervision.

Meta Planner Agent Onboarding

The onboarding steps are the same as the Meta Agent. See Meta Agent Onboarding:

  1. Select Template: Choose the Meta Planner Agent template.
  2. Select Agents: From the available list, select the worker agents to be managed by the Meta Planner Agent. The Meta Planner Agent supports binding of the following worker agent templates:
    • React Agent
    • React Critic Agent
    • Multi Agent
    • Planner Executor Agent
    • Planner Executor Critic Agent
  3. Agent Name: Provide a name for your meta planner agent.
  4. Agent Goal: Define the main objective.
  5. Workflow Description: Give instructions and guidelines for orchestration and task delegation.
  6. Model Name: Select the model for generating system prompts.

System Prompts

The Meta Planner Agent uses three system prompts:

  • Meta Planner System Prompt: Guides high-level planning and query decomposition.
  • Meta Responser System Prompt: Directs aggregation and synthesis of responses from worker agents.
  • Meta Supervisor System Prompt: Oversees execution, coordination, and quality control.

Agent Updation

To update the agent (add/remove worker agents, update workflow), follow the steps in Meta Agent Updation.

Agent Deletion

See Meta Agent Deletion for deletion steps and permission notes.

Important

Only the original creator of the agent can update or delete it. Other users cannot modify or remove these resources.