- dlib/integrations/paperless.py: sync HTTP client wrapping the paperless-ngx
REST API (upload doc, poll task, create/delete share links, delete document)
- config: PAPERLESS_ENABLED, PAPERLESS_BASE_URL, PAPERLESS_TOKEN, PAPERLESS_TAG_IDS
- PublicAsset model: paperless_document_id + paperless_share_slug columns
- publication service: after creating the asset, if paperless is enabled upload
the patched PDF and create a share link; stores doc id + share slug on the asset
- public routes: pass expires_at through to publish_version; new
POST /{slug}/share-links endpoint to (re)create expiring share links on demand
- schemas: PublishRequest.expires_at, PublicAssetResponse.paperless_share_url,
new ShareLinkRequest model
- frontend: paperless_share_url field on PublicAsset type, createShareLink()
and expiresAt param on publishVersion() in api.ts
- .env.example: documented paperless env vars
https://claude.ai/code/session_01YPVs6uBwCvcwVMvrfLBBdu
- dlib/ai/insights.py: pure-Python NLP analysis that correlates accepted
AI suggestion operations/keywords/sections with submission outcomes
(pending_review / published = positive, archived = negative)
- Backend: GET /api/v1/insights route + service + Pydantic schema
- Frontend: InsightsPanel component with bar charts for operation impact,
section impact, and keyword signal lift scores
- Insights tab added to the version panel; compact preview on doc overview
- NEXT_PUBLIC_DEMO=true makes the webapp fully standalone: loads
DEMO_DOCUMENTS / DEMO_SUBMISSIONS / DEMO_INSIGHTS from demo-data.ts,
disables all mutating actions, shows a DEMO badge in the top bar
- apps/webapp/public/demo-cv.docx: static dummy CV (Alex Rivera) for demo
- scripts/gen_demo_cv.py: script to regenerate the demo DOCX
- .env.example: document NEXT_PUBLIC_DEMO flag
https://claude.ai/code/session_01LWxu2qrwY6BRjUFXXn7NiM
- Replace ReportLab PDF export with LibreOffice headless for proper DOCX formatting preservation
- Add libreoffice-writer + fonts-liberation to backend Dockerfile
- Proxy public CV PDFs through frontend (/cv/[slug]) instead of redirecting to MinIO storage directly
- Fix docker-compose: route backend/worker to internal MinIO URL (http://cvfs-minio:9000), remove MinIO from public network, parameterize all domain/env vars
- Add storage cleanup (MinIO artifact deletion) when a document is deleted
- Add docker-compose.standalone.yml for self-deployment without Traefik/dokploy
- Update .env.example with comprehensive self-deployment documentation
https://claude.ai/code/session_017HGM9VPptZG52asT5pbL6Y
- documents.py: fix root_version_id never being saved due to SQLAlchemy
deferring the circular FK (CvDocument↔CvVersion). Use flush-based approach:
flush doc, flush version, then set root_version_id before final commit.
- config.py: add validator to coerce empty MINIO_ENDPOINT string to None,
preventing boto3 ValueError: Invalid endpoint at startup.
- versions.py: catch PatchValidationError and return 422 instead of 500
when a patch violates ATS guard rules.
- api.ts: on 401, clear stale OIDC cookies and redirect to /login instead
of showing the "Failed to load documents" error.
https://claude.ai/code/session_01KKbzWYz8fLyG2qcwiDZ8fy
- New monochrome design system in globals.css (no shadows, tight spacing, system-font)
- Root layout stripped to html/body; home page owns its Header/Footer
- Dashboard fully wired to FastAPI backend: upload, branch, submission, publish, download
- CVTree rebuilt as compact file-tree component driven by real version data
- DiffViewer rebuilt as minimal git-diff display using real patch data
- API client (libs/api.ts) extended with all CRUD operations
- Header redesigned to match minimal style
- Backend: added GET /documents/{id}/versions/{id}/download endpoint for DOCX export
https://claude.ai/code/session_01Xmxm2QLgFBgRJyYD6VukR6
