test-fix

Fix failing tests

Use when

• One or more tests fail and the cause is unknown.

Don't use when

• There are no tests yet (use test-write).
• The user wants a skip or deletion. Ask first and record why.

Steps

1. Run the failing tests and pipe the output through node <skill>/scripts/trim-test-output.mjs. Work from that summary; do not read the full log.
2. For each failure decide: is the code wrong, the test wrong, or the environment? Read the test and the code it calls. Say which, with the evidence.
3. Fix the code when the code is wrong. Fix the test only when its expectation is provably outdated or incorrect, and explain why.
4. Never remove an assertion, loosen a matcher, skip a test or raise a timeout to get green. If you think one is needed, stop and ask.
5. Re-run the same tests, then the wider suite.
6. Run git diff | node <skill>/scripts/weakened-assertions.mjs. Any finding must be shown to the user and approved.

Budget

• Search first (grep, glob, symbol search). Read line ranges, not whole files.

• Read at most 8 files per step. If you need more, write down what you learned and narrow the question.

• Do not re-read a file you just wrote or edited.

• Prefer the smallest command output that answers the question: head, tail, quiet flags, filtered lists.

• Trim output first; read only the failing tests and the code they call.

Output

• Per failure: cause, fix, and the file changed. Then the final pass counts and the weakened-assertions result.

Done when

• The failing tests pass and the wider suite has no new failures.
• weakened-assertions.mjs reports none, or the user approved each finding.