Source code for projects developed with Spec2Code framework

This dataset contains the complete source code for 18 production-ready mobile and web applications developed using the Spec2Code framework, an AI-assisted specification-driven development system. The applications span diverse domains including productivity tools, e-commerce platforms, educational systems, health and fitness trackers, and augmented reality applications. Each project includes: (1) complete specification documents generated from natural language descriptions, (2) implementation plans with 33 atomic tasks distributed across four development phases, (3) full source code with test suites achieving >85% coverage, (4) development logs documenting time expenditure and LLM interactions, and (5) architectural documentation detailing platform-specific adaptations (React Native, Firebase, Supabase, traditional backends).The dataset demonstrates AI-assisted development across varying complexity levels: simple CRUD applications (Recipe Finder, Finance Dashboard), medium-complexity applications with real-time features (Video Conference, Marketplace), and complex applications requiring advanced functionality (AR Decorator, LMS). Total development time across all 18 projects was 83 hours using Claude Sonnet 3.5, compared to estimated 1,560-2,140 hours for traditional development, representing a 19-26\u00d7 time reduction.This dataset enables research on: (1) AI-assisted software engineering productivity and quality metrics, (2) LLM behavior patterns in multi-phase development workflows, (3) specification-to-code generation effectiveness, (4) test coverage and code quality in AI-generated applications, (5) platform-specific architectural adaptation strategies, and (6) comparative analysis of AI development frameworks. All applications are functional, deployable, and include comprehensive test suites, making this dataset suitable for empirical studies on AI-assisted software development at scale.Keywords: AI-assisted development, LLM code generation, specification-driven development, mobile applications, test-driven development, software productivity