unity-test-runner_skill

Overview

This skill executes and analyzes Unity Test Framework tests from the command line. It detects Unity Editor installations, configures EditMode/PlayMode runs, executes tests via the Unity CLI, parses NUnit XML results, and produces actionable failure reports. Use it to automate test runs, surface file:line failures, and integrate Unity tests into CI pipelines.

How this skill works

The skill locates the Unity Editor across platforms and verifies the current directory is a Unity project. It builds a CLI test command with selected mode, platform, categories, and filters, runs Unity in batch mode, and captures test output. After execution it parses the NUnit XML, extracts summaries and failures, matches failures against a pattern database, and generates a detailed report with suggested fixes and code examples.

When to use it

• Run all Unity tests or a subset from the command line
• Validate game logic with EditMode and engine behavior with PlayMode tests
• Integrate Unity test runs into CI/CD pipelines for automated checks
• Investigate failing tests with file:line stack traces and categorized analysis
• Quickly re-run tests after code changes to verify fixes

Best practices

• Run EditMode tests first for fast feedback, then run PlayMode when needed
• Use test categories or filters to iterate quickly on specific features
• Set timeouts appropriate to test suites (shorter for EditMode, longer for PlayMode)
• Ensure Editor version matches ProjectVersion to avoid environment-specific failures
• Store and preserve XML test results for debugging and historical comparison
• Analyze grouped failure patterns and fix root causes instead of individual symptoms

Example use cases

• Execute all EditMode tests locally before merging a feature branch
• Run PlayMode Combat tests on a CI agent and fail the build if regressions appear
• Filter tests to the AI category while developing enemy behavior
• Parse test results to extract failing file paths and produce an actionable report
• Diagnose timeout or null-reference failures with suggested fixes and code snippets