Re-check implementation against re-derived user goals

Have an agent reconstruct the user’s requirements from all messages, then audit the code and tests against that list before changing anything.

Done with Custom agent· Intermediate· Oct 2026

What they did
A developer set up a custom agent to reread every message in a project and rewrite the user’s aim as a compact requirement list. They used that list as a benchmark for checking the code, test coverage, logs, and the live app. For larger tasks, the agent also maintained a project todo file and updated it as each review step was completed.
How it went
They say they have used this method for around six months, and that it has helped them catch fewer bugs and keep the work aligned with what the user asked for. No counts or other metrics were shared.
Tips
Keep the reconstructed requirements separate from the implementation so they can be checked on their own. Review the todo file both before and after the main change so gaps are easier to spot.

Try it yourself with Claude Code

Go through every message I’ve sent in this project and restate my goal as a brief checklist of requirements in your own words. Then inspect [folder or branch] against that checklist using the code, tests, logs, and the running app, and note any places where the implementation diverges, has no test coverage, or still needs attention. Write the findings into todo-review.md, then stop and ask for my approval before you edit code, remove files, or submit anything.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on news.ycombinator.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.