Planner and Executor Prompts for a Multi-Step Task
Separate planning from doing so an agent's steps can be reviewed.
Added October 15, 2026. Written with AI assistance and checked for clarity and placeholders; not reviewed by a lawyer, accountant, clinician or other professional. How these prompts are made
The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.
Write a pair of prompts for a planner and an executor handling [TASK_TYPE]. Available tools: [TOOLS]. Risks if a step goes wrong: [RISKS]. A typical input: [EXAMPLE_INPUT]. The planner prompt should produce a numbered plan where each step has an objective, the tool, the input, the expected output, a success check, and a rollback note, and should flag steps that need human approval. The executor prompt should run one step at a time, report the result against the check, stop on failure, and never alter the plan without returning to the planner. Provide the JSON format for the plan and the step result. Add rules for plans longer than eight steps (split them), for unknown inputs (ask), and for steps that repeat (limit retries to two). Show a worked example plan for the task and one failure with the agent's report. List what a reviewer should check in a plan before approving it and the common planning mistakes (vague steps, missing checks, hidden irreversible actions).
Fill in before use
Optional: type your details and the prompt above updates. Nothing is saved or sent.
Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.