6-Part Structure for Autonomous AI Agent Prompts

Use this 6-part structure for any AI tool to run autonomous tasks: 1. The Task One clear sentence saying what you want done. Not "check my code" but "run a full end-to-end QA test on the authentication flow." 2. The Resources Tell AI what it has to work with. How many parallel runs, what tools, what setup. 3. The Autonomy Level Define how much freedom AI gets. Can it make decisions on its own? Can it create files, run tests, push changes? Saying "do this autonomously" produces different output than saying nothing. 4. The Goal Give a specific, measurable target. "Find 200 bugs" is a goal. "Find bugs" is a wish. Numbers force the AI to keep working instead of stopping after the first 3 results. 5. The Guardrails Set quality standards AND boundaries. "Fix the root cause, no shortcuts" is a quality bar. "Don't touch the plugin layer" is a boundary. 6. The Output Tell AI where and how to deliver the result. "Write a report as markdown on my desktop and keep updating it" beats "let me know what you find" every time.