v1.0.0 — Initial Release

DocsChat v1.0.0

RAG chat system: ingest documents (PDF, DOCX, TXT, MD), embed them in ChromaDB, and chat with them using any LLM.

Features

• Document ingestion — PDF (pymupdf4llm), DOCX (python-docx), TXT, Markdown
• Recursive text chunking — Custom splitter with configurable size and overlap
• Pluggable embeddings — OpenAI (text-embedding-3-small) or local (sentence-transformers)
• Multi-LLM support — OpenAI, Gemini, Ollama with messages-based API
• RAG pipeline — Vector search → context building → multi-turn chat with query rephrasing
• FastAPI REST API — Full CRUD for collections and conversations, SSE streaming
• Typer CLI — Interactive REPL with Rich rendering, single-question mode
• SQLite database — Async (aiosqlite) for conversations, messages, and document metadata
• Optional API key auth — Middleware for securing API endpoints
• CI/CD — GitHub Actions with lint + test matrix (Python 3.10-3.12)