Make AI Coding Requests Reviewable: Five Habits and an Acceptance Checklist

说明与用途

将模糊的 AI 编码请求转化为结构化、可验证的需求，简化评审并降低错误率。

问题

AI‑assisted code requests often mix multiple requirements, making them hard to verify and review.

结果

The workflow transforms vague AI requests into clear, testable requirements, improving reviewability and reducing bugs in the final code.

工作流

• 1. Identify a hard‑to‑verify request.
• 2. Number each independently checkable behavior (R1, R2, …).
• 3. Add short clarifications in parentheses to specify targets.
• 4. State the overall goal and current understanding, asking the assistant to verify assumptions.
• 5. Restate current and intended behavior on every revision.
• 6. Separate combined requests into individual rules while preserving logical conjunctions.
• 7. Produce a reviewable replacement prompt with numbered requirements and verification steps.
• 8. Translate requirements into observable checks (input → expected observation).
• 9. Keep subsequent revisions explicit and update baselines as needed.
• 10. Review AI‑generated changes item by item, marking verified, incorrect, or unverified items.

证据

Give every independently checkable behavior its own item. This provides a stable reference for implementation and review: “R3 is missing” is more useful than “the form still feels wrong.”

来源

For the feature being changed, say what it does now and what it should do afterward. Include behavior that must be preserved.

来源