cc-check

Reference Files

• test-strategies.md - Type-specific test execution strategies
• report-template.md - Complete output format template
• examples.md - Concrete test cases; reference when creating tests
• common-failures.md - Failure patterns and fixes; reference when debugging

Focus Areas

• Sample Query Generation - Creating realistic test queries based on descriptions
• Expected Behavior Validation - Verifying outputs match specifications
• Regression Testing - Ensuring changes don't break existing functionality
• Edge Case Identification - Finding unusual scenarios and boundary conditions
• Integration Testing - Validating customizations work together
• Performance Assessment - Analyzing context usage and efficiency

Test Framework

Test Types

Test Execution Strategy

Type-specific strategies in test-strategies.md.

Test Process

Step 1: Identify Customization Type

Determine what to test:

• Agent (in agents/)
• Skill (in skills/)
• Hook (in hooks/)

Step 2: Read Documentation

Use Read tool to examine:

• Primary file content
• Frontmatter/configuration
• Usage instructions
• Examples (if provided)

Step 3: Generate Test Cases

Based on description and documentation:

For Skills: Extract trigger phrases, create 5-10 sample queries that should trigger, 3-5 that should NOT trigger

For Agents: Create prompts based on focus areas, scenarios inside and outside scope

For Hooks: Create inputs that should pass, block, and malformed inputs

Step 4: Execute Tests

Choose testing mode based on what you need to verify:

Read-Only Testing (default): Analyze configurations and documentation to assess expected behavior

Active Testing (when appropriate): Actually invoke skills with sample queries or trigger hooks

Test priority: Always run discovery + frontmatter + functional tests. Add integration tests only when the target references or interacts with other customizations.

Step 5: Compare Results

For each test:

• Expected: What should happen (from docs/description)
• Actual: What did happen (from testing)
• Status: PASS (matched) / FAIL (didn't match) / EDGE CASE (unexpected)

Step 6: Generate Test Report

Create structured report following report-template.md.

Output Format

Test reports follow a structured format. See report-template.md for the complete template.

Key sections: Summary, Test Results, Functional Tests, Integration Tests, Usability Assessment, Edge Cases, Recommendations

Best Practices for Testing

1. Test Both Paths: Success cases AND failure cases
2. Edge Cases Matter: Test boundaries and unusual inputs
3. Clear Expected Behavior: Document what should happen
4. Realistic Queries: Use natural language users would actually type
5. Integration Testing: Don't just test in isolation
6. Performance Aware: Note if tests are slow or heavy
7. Regression Testing: Re-test after changes
8. Document Failures: Explain why tests failed, not just that they did
9. Actionable Recommendations: Provide specific fixes
10. Version Testing: Note which version was tested

Common Test Failures

For failure patterns and fixes by customization type, see common-failures.md.

Tools Used

• Read - Examine customization files
• Write - Generate and save test reports
• Grep - Search for patterns and configurations
• Glob - Find files and customizations
• Bash - Execute read-only commands for analysis
• Skill - Invoke skills for active testing

Test reports are output inline by default. To persist, write to ~/.claude/logs/evaluations/tests/ (create the directory if needed with mkdir -p).

Related Skills