Planning Agent
An AI agent specialized in decomposing high-level goals into ordered, executable sub-tasks, selecting tools and sub-agents, and maintaining a plan that adapts as new information arrives during execution.
Planning agents operate at the top of the agentic hierarchy. Given a natural-language objective, a planning agent produces a structured task graph—a sequence or DAG of steps—then either executes them directly or delegates to specialized sub-agents. Common planning strategies include tree-of-thought search, ReAct-style planning, and divide-and-conquer decomposition.
Good planning agents are also re-planning agents: when a step fails or produces unexpected output, they revise the remaining plan rather than aborting. This requires tight integration with an episodic memory that records what has already been tried.
In systems like Paperclip, the planning role is distinct from the execution role, enforcing the principle that a single agent should not both design and execute work without review. The planner produces a plan artifact that a human or senior agent can inspect before execution begins.
## Related Terms
- [[glossary/agent-loop|Agent Loop]] — the iterative perceive-act-observe cycle the harness executes - [[glossary/hierarchical-agents|Hierarchical Agents]] — the pattern of orchestrator-plus-specialist agents with layered authority - [[glossary/sub-agent|Sub-Agent]] — a child agent spawned by an orchestrator to execute a bounded subtask - [[glossary/orchestrator|Orchestrator]] — the component that owns the task graph and decides which agent handles each step - [[courses/claude-agent-sdk-zero-to-production|Course: Claude Agent SDK — Zero to Production]] — hands-on practice with the concepts covered in this entry