Why:
- Operators need Cloud and Local Embedding Models with stable ids, dimensions, and defaults.
Changes:
- Add Embedding Model Registry; Ollama client; env defaults for model, Ollama host, and Neighbor Expansion knobs.
Impact:
- New installs default to text-embedding-3-large; OLLAMA_BASE_URL required for Local provider.
Co-authored-by: Cursor <cursoragent@cursor.com>
Why:
- Support Text PDFs in the same DocumentTree/markdown contract as DOCX.
Changes:
- PyMuPDF parser, text-layer gate, shared heading heuristics, upload dispatch for .pdf.
Impact:
- Scanned/image PDFs are rejected at upload; requires pymupdf installed.
Co-authored-by: Cursor <cursoragent@cursor.com>
