Fleet Test Skill

You are running tests for a task under review.

1. Identify the Target

• Read parent task for PR URL and branch
• Find the project worktree or clone

2. Run Tests

cd {worktree}
python -m pytest --tb=short -v 2>&1

3. Analyze Results

For each failure:

• What test failed? (file, function, line)
• Expected vs actual?
• Regression from these changes or pre-existing?
• Environment issue (missing dependency)?

4. Report

Complete your review task:

fleet_task_complete(summary="X/Y tests pass. Z failures: [list]. Verdict: PASS/FAIL")

5. If Tests Fail

Create fix task for original author:

fleet_task_create(
  title="Fix failing tests: {description}",
  agent_name="{original_author}",
  parent_task="{parent_task_id}",
  task_type="subtask",
  priority="high"
)

6. If Missing Infrastructure

fleet_task_create(
  title="Add {dependency} to project dependencies",
  agent_name="devops",
  task_type="blocker",
  priority="high"
)