---
name: fix-test
description: "Auto-diagnose and fix failing tests — up to 3 attempts with re-runs. Triggers on: fix test, failing test, broken test, test error, why is this test failing, debug test."
---
# Fix Test

**IMPORTANT: Start your response with a context preamble.**

Call `help_lookup(topic="fix-test", mode="preamble")` and
display the returned `preamble` text as a blockquote. Then
tell the user they can say "tell me more" for a step-by-step
guide, or answer the scoping questions below to proceed.

If the MCP call fails, fall back to:

> **Fix Test** — Diagnoses failing tests, classifies the root cause, and applies targeted fixes automatically.

## Scoping

Before running, ask:

1. **Target**: "Which test is failing? A specific file,
   test name, or should I find failures automatically?"
2. **Context**: "Did this start failing after a recent
   change, or has it been broken?"

## Execution

### Step 1: Identify Failures

Run the failing test(s) to capture the error:

```bash
uv run pytest <target> -v --tb=short 2>&1 | tail -40
```

### Step 2: Diagnose Root Cause

Common failure patterns:

| Pattern | Root Cause | Fix |
| ------- | ---------- | --- |
| `ModuleNotFoundError` | Import path changed | Update import |
| `AttributeError: mock` | Mock target wrong | Match import path |
| `AssertionError` | Expected value drift | Update assertion |
| `TypeError: __init__` | Constructor changed | Update call site |
| `FileNotFoundError` | Fixture path wrong | Use `tmp_path` |

### Step 3: Apply Fix and Re-run

Apply the fix, then re-run the test. If it still fails,
diagnose again with the new error. Repeat up to 3 times.

```bash
uv run pytest <target> -v --tb=short
```

### Step 4: Report

After fixing (or exhausting 3 attempts), report:

```markdown
## Fix Test Results

**Tests Fixed:** X/Y | **Attempts Used:** Z/3

### Fixed
| Test | Root Cause | Fix Applied |
|------|------------|-------------|

### Still Failing (if any)
| Test | Error | Attempts | Notes |
|------|-------|----------|-------|
```

## Follow-Up

After presenting results, offer:

- "Want me to generate missing tests for the fixed
  module?"
- "Should I check for similar failures elsewhere?"
- "Want a deeper look at the root cause?"