v1.0.5

🎉 SemanticTest v1.0.5

🐛 Bug Fixes

Fixed undefined path resolution in ValidateContent

• Fixed critical bug where undefined values in from/as input were treated as literal strings
• ValidateContent now properly handles missing data paths and returns validation failures
• Changed core path resolution from resolveValue() to dataBus.get() for accurate undefined handling

Enhanced array contains operator

• Arrays now support substring matching: checks if any element contains OR equals the expected value
• Improves flexibility when validating arrays of strings

📚 Documentation

Complete example documentation (4 new pages)

• ✅ Basic API Test - REST API testing patterns
• ✅ AI Chat Test - LLMJudge semantic validation
• ✅ Calendar Agent - Tool validation patterns
• ✅ Error Handling - Error handling patterns

Updated links and references

• Documentation now available at docs.semantictest.dev
• Main website at semantictest.dev
• Updated all README references to new URLs

🗂️ Examples

Improved example structure

• Unified all examples into single test-examples/ directory
• All examples now use context variables (${BASE_URL}, ${USER_ID}) for better reusability
• Renamed examples for clarity:
  • test-error-reporting.json → validation-test.json
  • test-reporting.json → mock-ai-validation.json
• Removed environment-dependent examples for better portability

✅ Testing

• 453 tests passing (420 unit + 23 integration + 10 examples)
• 100% pass rate across all test suites
• All documentation examples tested and verified

📦 Installation

npm install @blade47/semantic-test@latest

🔗 Links

- 📖 Documentation: https://docs.semantictest.dev
- 🌐 Website: https://semantictest.dev
- 💻 GitHub: https://github.com/blade47/semantic-test
- 📝 Examples: See test-examples/ directory

v1.0.4

v1.0.4 - CLI Fixes

🐛 Bug Fixes

Fixed CLI execution issues - The npx semtest command now works correctly in all scenarios:

• Fixed dynamic import for OpenAI - The CLI no longer crashes when the openai package is not installed. LLMJudge now uses lazy loading, allowing
  users to run tests without LLM evaluation without needing to install OpenAI.
• Fixed npx execution via symlinks - Resolved issue where npx semtest produced no output due to symlink resolution problems in
  node_modules/.bin/. The CLI now properly detects when it's run directly vs imported as a module.

📦 Installation & Usage

Install locally

npm install @blade47/semantic-test

Use with npx (locally installed)

npx semtest your-test.json

Or run directly without installing

npx @blade47/semantic-test@latest your-test.json

🔧 Technical Details

• Dynamic OpenAI import only loads when LLMJudge is actually used
• Added fs.realpathSync() to resolve symlinks when detecting direct script execution
• Improved error messages for missing OpenAI package

✅ Verified

• All 359 tests passing
• Works with local installation (npm install)
• Works with global execution (npx @blade47/semantic-test)
• Proper error handling for optional peer dependencies

v1.0.1

Package moved to private npm namespace and complete renaming from old branding.

v1.0.0

Initial release - Semantic testing for AI systems