Universal Test Runner

Auto-detect the project type and run the appropriate test suite. Works across Python, Node.js, and Go projects.

Detection Logic

Detect the project type by checking for manifest files in the current working directory. Check in this order (first match wins):

1. Go: go.mod exists
2. Python: pytest.ini, pyproject.toml (with [tool.pytest]), or setup.py exists
3. Node.js: package.json exists — then check:
   • If vitest in devDependencies -> vitest project
   • If jest in devDependencies -> jest project
   • If test script defined in scripts.test -> use that

Also detect the package manager for Node.js projects:

• pnpm-lock.yaml exists -> pnpm
• yarn.lock exists -> yarn
• bun.lockb exists -> bun
• Otherwise -> npm

Arguments

Parse $ARGUMENTS for flags and positional arguments:

• No arguments: Run the default test suite
• --unit: Run only unit tests
• --integration: Run only integration tests
• --e2e: Run only E2E tests
• --coverage: Run with coverage reporting
• <path>: Run tests for a specific file or directory
• Flags can be combined: /test --unit --coverage

Test Commands by Project Type

Python (pytest)

Detection refinements for Python:

1. Check if tests/unit/ and tests/integration/ directories exist
2. If they exist, use directory-based scoping
3. If not, check pytest.ini or pyproject.toml for markers and use -m flag
4. If the project uses markers, report which markers are available

Node.js (vitest/jest)

Detection refinements for Node.js:

1. Read package.json scripts section for existing test commands (test:unit, test:e2e, test:integration, test:coverage)
2. If specific scripts exist, prefer those over generic flags
3. Check for Playwright config (playwright.config.ts/.js) for E2E
4. Report available test scripts from package.json

Go

Execution

1. Detect the project type and report it:

   Detected: Python/pytest project

2. Build the command based on arguments
3. Check for project-specific overrides in CLAUDE.md:
   • Look for a "Testing" or "Test" section
   • If CLAUDE.md specifies custom test commands, prefer those
   • Report: "Using custom test command from CLAUDE.md" if applicable
4. Run the test command using Bash tool
   • For long-running test suites (e2e, full integration), suggest tmux but run directly
   • Stream output so the user can see progress
5. Report results:

   Tests: X passed, Y failed, Z skipped
   Coverage: N% (if --coverage)
   Duration: Xs

Error Handling

• If no manifest files found: report "Could not detect project type. No manifest files found in current directory."
• If test directory doesn't exist for the requested scope: report which scopes are available
• If tests fail: show the failure summary but don't treat it as a skill error

Notes

• This skill wraps existing project test infrastructure — it does not install test frameworks
• For projects with both Python and Node.js tests (e.g., a monorepo with a Python backend and a JS frontend), detect both and ask which to run, or run both if no scope flag is given
• Always respect the project's existing test configuration (pytest.ini, vitest.config.ts, etc.)