0.1

Initial release

• On-Device Vector Embedding Generation: Generates 768-dimensional text embeddings using MLX framework with support for multiple pre-trained models (nomic_text_v1_5, bge_small, etc.)
• Persistent JSON-Based Storage: Implements file-based document storage with automatic serialization/deserialization of vector embeddings and metadata
• Cosine Similarity Search: Performs vector similarity calculations with configurable threshold (default: 0.7) and result limits
• BM25 Text Indexing: Implements term frequency-inverse document frequency algorithm for keyword-based document retrieval
• Asynchronous API: Built with Swift's async/await concurrency model for non-blocking operations
• Batch Processing: Supports parallel embedding generation for multiple documents with automatic UUID assignment
• Modular Storage Protocol: Abstracts storage implementation through protocol-based design for extensibility
• Command-Line Interface: Provides CRUD operations through ArgumentParser-based CLI with database management commands