PHPUnit Fixer Skill

This skill analyzes EXISTING PHPUnit test failure logs and implements fixes. It does NOT run tests.

Agent Delegation Strategy

This skill delegates to the php-qa-ci_phpunit-fixer agent (sonnet model).

Workflow

When User Says: "Fix the test failures"

1. Launch fixer agent:

   Use Task tool:
     description: "Fix PHPUnit test failures"
     subagent_type: "php-qa-ci_phpunit-fixer"
     prompt: "Find and fix failures in most recent test log"
   

2. Receive fixer output with:

   • Errors found and grouped by pattern
   • Fixes applied
   • Files modified

3. If no log found:

   • Suggest using phpunit-runner skill to generate log first

When User Provides Specific Log Path

1. Launch fixer agent with explicit log path:

   Use Task tool:
     description: "Fix test failures from log"
     subagent_type: "php-qa-ci_phpunit-fixer"
     prompt: "Fix failures in log: {user_provided_path}"
   

Escalation Triggers

Launch opus model or ask human when:

• Fixer agent reports business logic questions (test vs code expectations)
• Same error pattern persists after 2 fix attempts
• User asks for explanation rather than fixes

Fixer Agent Reference

The phpunit-fixer agent (sonnet model) handles:

• Auto-discovery of most recent JUnit XML log
• Error parsing and pattern grouping
• Fix implementation for common patterns
• Reporting which files were changed

See .claude/agents/php-qa-ci_phpunit-fixer.md for agent implementation details.

When to Use This Skill vs phpunit-runner

• Use phpunit-fixer when:

  • Tests were already run manually
  • You have a specific log file to analyze
  • You only want to analyze/fix, not run tests

• Use phpunit-runner when:

  • You want to run tests AND fix failures
  • You want the full run→fix→run cycle
  • Tests haven't been run yet