Agent completion receipt prompt.
When an AI agent says "done," make it hand you a receipt. Not a victory lap. A receipt: what changed, how it was checked, and what is still not proven.
Copy this before accepting the work.
Before you call this done, give me a completion receipt. Use plain English. Include: 1. What changed. 2. Why it matters. 3. The exact proof I can inspect. 4. What you verified yourself. 5. What is still unproven. 6. The one next step if I want stronger proof. Do not summarize confidence. Show evidence. If the original goal is not fully proven, say that clearly.
The trap this avoids
An agent can make a button open, a script run, or a page render and still miss the real goal. The receipt forces the agent to match the proof to the claim.
What a useful receipt names.
- The exact public URL, file path, command output, screenshot, or dashboard item.
- The difference between local testing and the live surface a real user sees.
- The difference between opening checkout and proving a real payment happened.
- The one thing still missing if the goal is not truly finished.
Use the full checklist for higher-stakes work.
The $1 download includes this receipt prompt, the one-minute handoff brief, 15 acceptance checks, proof menus, rejection flags, and follow-up lines for fuzzy agent answers.
Use the free generator first if you are still writing the task. Buy the checklist when you are reviewing the finished work.
Buy the full checklist for $1