Guides

guides on reviewing coding agents' work.

Practical notes on checking what coding agents write: what breaks, and how to catch it. Written by Lay Sheth.

  1. Your coding agent said done. Here's how to check.

    Claude Code says done but the tests fail? The five false dones, how to catch each one with git and a fresh test run, and why the summary isn't proof.

  2. plottypus is open: a free trial for Claude Code

    Anyone can now try plottypus, live review of a Claude Code session that catches false dones and messages the agent with proof. No card needed.

  3. A checklist for reviewing a coding agent's work.

    How to review code a coding agent wrote: read the ask first, check test edits on their own, rerun the tests, watch scope. Copy-paste checklist included.

  4. Should one coding agent review another's work?

    Yes, if the reviewer is not the agent that wrote the code and it checks evidence, not the summary. What a reviewer catches, and how to set one up by hand.

  5. Make your coding agent rerun the tests before it says done.

    A Claude Code Stop hook that reruns the tests and blocks the turn while they fail, checked against the docs, with its limits and a trusted test copy.

  6. What happens when two coding agents edit the same file?

    One of them loses, often quietly. What two studies found, why agents from different vendors clash twice as often, and five ways to cut it down.

tell us your case.

When did a coding agent last tell you it was done, and it wasn't? A person reads every answer.