AI-powered test generation for C#, VB.NET, React, Vue.js, T-SQL with self-learning patterns. Generate complete test suites in minutes, not days.
Trusted by developers worldwide
From frontend to backend to database, generate comprehensive tests across your entire stack
xUnit, NUnit, MSTest
Legacy Modernization
Jest + Testing Library
Vitest + Test Utils
tSQLt Framework
Playwright E2E
All languages, one command:
python generate_tests_v2.py /project --language <your-stack>
Not just template-based generation. Our AI learns your project patterns and improves with every run.
Discovers 12+ project-specific patterns in 3 minutes by analyzing compilation errors. Learns property names, return types, enum values, and type hints automatically.
Zero manual intervention required. The pattern learner compiles your tests, analyzes errors, and automatically caches patterns for future runs.
Choose from OpenAI (GPT-4o mini/4o), Anthropic Claude (3.5 Haiku), or Google Gemini (2.0 Flash). Default uses GPT-4o mini at $0.04/test - 99% cheaper than GPT-4 Turbo.
Not just placeholder code. Generates comprehensive tests with proper mocking, async/await support, exception handling, and framework best practices.
Three simple steps to comprehensive test coverage
Point the tool at your codebase. It auto-detects language, parses classes/components, and extracts methods with dependencies.
AI analyzes each file and generates comprehensive tests using framework best practices and proper mocking patterns.
Pattern learner analyzes compilation errors and caches corrections. Next run uses learned patterns with 30-40% fewer errors.
Total time for 44 files: 8 minutes
See it in actionPay only for what you use. Choose your AI provider based on speed and cost needs.
Best value for most projects
Fast & cost-effective
Google's latest model
All prices include pattern learning cost (~$0.50 one-time per project)
Switch providers anytime with --provider <openai|claude|gemini>
See how much time and money you'll save with AI-powered test generation
Production validation from the RemoteC Enterprise project
ASP.NET Core 8.0 WebRTC Platform • 15,000+ LOC
Fully autonomous workflow: Zero manual intervention during generation and pattern learning phases
Cost efficiency: Total $2.23 vs estimated $1,467 for manual testing (99.8% savings)
Time savings: 8 minutes vs 14.7 hours manual effort (110x faster)
Self-learning validation: Documented 266 baseline errors provide clear benchmark for measuring 30-40% improvement on next run
Watch how easy it is to generate comprehensive test suites
git clone https://github.com/\ndbbuilder/dotnet-unit-test-gen\ncd dotnet-unit-test-gen\npip install -r requirements.txt
echo "OPENAI_API_KEY=sk-..." > .env\n# Or use ANTHROPIC_API_KEY\n# Or use GOOGLE_API_KEY
python generate_tests_v2.py \\\n /path/to/project \\\n --language csharp
The generated tests follow framework best practices and include proper mocking, async/await support, and comprehensive coverage. While you may need to adjust some tests for project-specific edge cases, the RemoteC case study showed that with pattern learning, 60-70% of tests work out of the box, and most errors are minor (e.g., property names, enum values).
We recommend GPT-4o mini (default) for the best balance of cost, speed, and quality. It's 99% cheaper than GPT-4 Turbo while maintaining excellent output quality. Claude 3.5 Haiku is faster but slightly more expensive. Gemini 2.0 Flash is the cheapest option but may require more manual corrections.
After generating tests, you can run the pattern learner (using LangChain 1.0) which compiles your tests, analyzes compilation errors, and extracts project-specific patterns like property name corrections, return type mappings, and enum values. These patterns are cached in .test-gen-cache/ and automatically applied in future runs, reducing errors by 30-40%.
Yes! v1.3 includes full support for React (Jest + React Testing Library) and Vue.js 3 (Vitest + Test Utils). The tool handles components, hooks/composables, state management, and follows accessibility best practices. You can even generate integration tests that span frontend, backend, and database layers.
The T-SQL handler generates tSQLt tests for stored procedures, functions, and views. It includes FakeTable mocking for database dependencies and proper transaction handling. This is perfect for legacy databases that need test coverage before modernization.
Yes! The tool was validated on a production enterprise project (RemoteC) with 44 test files and 15,000+ lines of code. It successfully handled complex ASP.NET Core 8.0 code including dependency injection, async operations, and custom business logic. The pattern learning system has been tested with real-world compilation errors and produces measurable improvements.
C#/VB.NET support xUnit (default), NUnit, and MSTest. React uses Jest + React Testing Library. Vue.js uses Vitest + Vue Test Utils. T-SQL uses tSQLt. Integration tests use Playwright for E2E testing. All frameworks follow industry best practices and official documentation.
Absolutely! The tool is open-source (GitHub: dbbuilder/dotnet-unit-test-gen). The modular architecture makes it easy to add new language handlers, test frameworks, or AI providers. Contributions are welcome - whether it's bug fixes, new features, or documentation improvements.
Join developers worldwide who are saving hours every week with AI-powered test generation. Free, open-source, and production-ready.