Test Review

Three defects, in the order they matter:

1. Tautology — the assertion cannot fail, so the test is green forever.
2. Incompleteness — the assertion can fail, but passes over regressions a stronger one would catch.
3. Coupling — the assertion is strong and meaningful, but written against internals, so it breaks every time the implementation is improved.

The first two decide whether a test catches bugs. The third decides whether the codebase stays safe to change. Fix a tautology before bothering to strengthen a subset assertion — there is no point tightening an assertion that was never load-bearing.

Target

Scope comes from $ARGUMENTS: a path, commit, range, branch, or PR number; empty means uncommitted changes plus the current branch's diff vs its base. For a single commit, compare against its first parent. --fix anywhere in the args applies findings after reporting; otherwise report only.

Review test files only — *.spec.*, *.test.*, __tests__/**, and the project's equivalents. Never flag production code. Report only tests the change added or touched; a pre-existing test elsewhere is out of scope, and when one genuinely blocks reading a new one, say so as a note.

1. Tautology

Flag any assertion whose outcome does not depend on the code under test. A tautological test is worse than no test: it manufactures confidence and survives real regressions untouched.