Eval-driven tool design

Trace-detective lab: a file search tool keeps failing because its description says one thing and the tool does another. Read the failure transcripts, name the cause, rewrite the description, and choose how the rewrite is judged before it ships: fresh tasks set aside first, each checked for exactly the right files. Catch a draft guard that passed a description forbidding the fix, and read what the published post does and does not say.

This Harness Engineering lesson is part of a paid plan. Sign in with a plan that includes this course to read it.