diagnose-test

name	diagnose-test
description	Debug a failing test by loading full context — test file, source file, error output
argument-hint	<test-file-or-pattern>
allowed-tools	Bash(bun test *) Read Glob Grep

Diagnose Failing Test

Debug a failing vlist test with full context.

Instructions

1. Run the failing test:

   • bun test $ARGUMENTS 2>&1

2. If the test passes, report that and stop.

3. If it fails, gather context:

   • Read the full test file
   • Identify which source file(s) the test covers (from imports)
   • Read those source files
   • Read relevant test helpers if used (test/helpers/)

4. Analyze the failure:

   • What does the test expect?
   • What actually happened?
   • Is this a test bug or a source bug?
   • If it's a recent regression, check git log --oneline -10 -- <source-file> for recent changes

5. Output:

   ## Failing Test
   **File:** test/plugins/async/plugin.test.ts
   **Test:** "should cancel load when velocity exceeds threshold"
   **Error:** Expected 0 but received 1
   
   ## Diagnosis
   The test expects `loadAdapter.load` to not be called when velocity > LOAD_VELOCITY_THRESHOLD,
   but the velocity check in src/features/async/feature.ts:87 uses `>=` instead of `>`.
   
   ## Suggested Fix
   Change line 87 from `velocity >= threshold` to `velocity > threshold`,
   or update the test to use a velocity strictly above the threshold.
   

6. Be specific — include line numbers, variable names, and exact values.