Add max_tokens parameter to NexAGI endpoint

Add privacy column to Document model + migration 005

Remove PII from research mode and response handlers

Tune Presidio: strict mode masks names+orgs for NDA, light mode skips them for research

Add privacy flag to upload — PII runs only when privacy=True (NDA mode)

Fix tuple unpacking error in gather result for PDF parsing

Skip MarkItDown for PDFs — run only as fallback if pypdf returns empty

Swap pdfplumber → pypdf for upload parsing (3x faster text extraction)

Fix NexAGI: reasoning_details is already a list of dicts

Fix NexAGI: reasoning_details list model_dump, add reasoning_details to message schema

Add /NexAGI endpoint — OpenRouter reasoning via nex-agi/nex-n2-pro:free

feat: provider-based LLM profiles + per-request model switching

fix: remove stale page_texts_pdf references from PII block

fix: use pdfplumber text for chunking to ensure accurate page numbers for citations

fix: keep MarkItDown for chunking, pass pdfplumber page texts separately so page numbers align for bbox extraction

fix: use pdfplumber text for chunking so page numbers align with page ranges (fixes bbox extraction)

fix: actually anonymize PII in document text before indexing + sanitize all LLM outputs

fix: populate username from GitHub login on OAuth signup, return email/username in OAuth login response

feat: add username field to User model, schemas, and API

perf: move PII detection from upload HTTP handler to job worker (runs concurrent with parsing)

feat: harden prompt injection guardrails with 11 detection layers + fix missing guardrails on /analyze

fix: add PII scan result log to upload endpoint

perf: revert to all-MiniLM-L6-v2, query embed cache, fire-and-forget Cloudinary, pre-warm reranker, reduce top-k

fix: import parse_document outside .pdf scope; coerce datetime→str in all response schemas

production hardening: circuit breakers, graceful degradation, health checks, parallel uploads, HF Spaces deploy config

fix: remove local file fallback, use Cloudinary only, add POST /documents/backfill endpoint for old docs

fix: add file_url to every citation so the frontend PDF viewer gets the URL directly without extra API call

fix: cloudinary_url now falls back to local file_url so frontend PDF viewer always gets a working link

fix: add file_url to DocumentResponse, pass request to _doc_response, handle old-path files in serve_file

fix: Cloudinary upload resource_type=auto for PDF preview instead of raw download

fix: Cloudinary upload_bytes->upload method name bug, add GET /files/{doc_id} endpoint for PDF serving fallback

fix: enable NeMo Guardrails with correct parameters.api_key/parameters.base_url, add PII detection to guardrails flow, set PII_ENABLED=true

fix: handle LLM connection errors gracefully (SSL/connect errors → 502); convert Analyze 500 → 502 with clear message

fix: auto-fallback to Qdrant :memory: when remote unavailable

fix: load .env at module level in main.py (removes run.py dependency); auto-start worker + register processor on startup

fix: upload pipeline completely dead - worker never started, processor never registered

fix: 15 critical/high bugs across codebase

fix upload pipeline: await add_document, pre-load model eagerly, add verbose UPLOAD/VectorStore logging; remove unused deployment configs

fix models.py: match Neon schema varchar sizes, add email unique constraint