Prefer the real mutation tool, and make a hand-rolled runner prove it ran - #8
Merged
Merged
Conversation
… ran The manual mutation procedure reads as an equal alternative to a mutation tool. It is not, and this repo's own six-round verification is the evidence: the hand-rolled runner reported kills for mutants it never executed, because two same-size mutants written in the same second shared a bytecode cache. What makes that failure worth a rule rather than a bugfix is its direction. It could only ever inflate the score. A mutant that silently does not run reports a kill, so the layer stays green precisely because it is broken, and no red gauntlet can surface it. Every other layer here fails loudly when it breaks; this one did not. Two changes follow. The SKILL.md cell now states a preference order rather than an equivalence: the project's tool generates mutants from the syntax tree and cannot skip one silently, so reach for it first. And the manual procedure now carries the execution-proof requirement, pointing at the guard tools/mutants.py already gained in the fix, so the next person writing a runner from this procedure knows it is required rather than incidental. The general form: a hand-written mutant list matched against source text is a second copy of the code, and it goes stale on every refactor of the thing it is meant to guard. (Our fork goes further and reports the layer UNAVAILABLE rather than falling back to a hand-rolled runner at all. That is a bigger change to this skill's posture, so it is not in this PR.) Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
drmikecrowe
marked this pull request as ready for review
August 12, 2026 15:31
This was referenced Aug 14, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The manual mutation procedure currently reads as an equal alternative to a mutation tool. It isn't — and this repo's own six-round verification is the evidence.
The hand-rolled runner reported kills for mutants it never executed, because two same-size mutants written in the same second shared a bytecode cache. (Fixed in #4/#5;
tools/mutants.pynow pins mtimes and aborts if the cache reappears.)What makes that worth a rule rather than just a bugfix is its direction. It could only ever inflate the score. A mutant that silently doesn't run reports a kill, so the layer stays green precisely because it is broken, and no red gauntlet can surface it. Every other layer in this skill fails loudly when it breaks. This one didn't.
Two changes:
SKILL.mdmutation cell states a preference order rather than an equivalence. A real tool generates mutants from the syntax tree: it can't apply a mutant to code that moved, and it can't report one it didn't run. Reach for it first; manual mutation is for languages with no tool.tools/mutants.pyalready gained — so the next person writing a runner from this procedure knows it's required rather than incidental, and EVIDENCE says which check proves execution.The general form, which is the part I'd argue for keeping even if you reword the rest: a hand-written mutant list matched against source text is a second copy of the code, and it goes stale on every refactor of the thing it's meant to guard.
+18/-1 across
SKILL.mdandreferences/gauntlet.md. Demo and CI untouched.Part of a series from a fork that has been running this skill daily (drmikecrowe/old-coder). Independent of the other PRs.
🤖 Generated with Claude Code