MCP RAG with ChromaDB

# Core dependencies langchain-chroma>=0.1.0 langchain-core>=0.1.0 langchain-text-splitters>=0.1.0 requests>=2.31.0 fastmcp>=0.1.0 # Note: We use direct Ollama API calls instead of langchain-ollama # to avoid port conflicts when running through MCP/Claude Desktop # Document format support PyPDF2>=3.0.0 # PDF support python-docx>=1.0.0 # DOCX support beautifulsoup4>=4.12.0 # HTML/XML support lxml>=4.9.0 # Better XML/HTML parsing python-pptx>=0.6.21 # PowerPoint support openpyxl>=3.1.0 # Excel support odfpy>=1.4.1 # ODT (OpenDocument) support markdown>=3.4.0 # Markdown support (optional enhancement) # Optional: Better PDF extraction (uncomment if needed) # pdfplumber>=0.10.0 # Alternative PDF extraction with better table support # pypdf>=3.0.0 # Modern PyPDF alternative # Optional: Additional format support # python-magic>=0.4.27 # File type detection # chardet>=5.0.0 # Character encoding detection # textract>=1.6.5 # Universal text extraction (requires system dependencies)