epic: full extraction and quality gate

[Fluory]

All fields incl. line items (product, quantity, material, dimensions), XLSX/DOCX/MSG and scanned PDFs, eval runner with 15 weighted synthetic cases and a CI regression gate. Milestone M2 (docs/product/roadmap.md).

Issues (in order)

• feat(ai-service): full field set including line items #22
• feat(ai-service): XLSX, DOCX, MSG and scanned PDFs #23
• feat(evals): eval runner, 15 weighted synthetic cases and CI gate #24
• feat(review): line items and all formats in the review UI #25

[Fluory]

Overnight run – status for this epic

All issues of this epic have a ready-for-review PR (stacked, not merged):

Issue	PR	State
#22 full field set	#39	ready, CI green
#23 formats (XLSX/DOCX/MSG/OCR)	#40	ready, CI green, security review done
#24 evals + CI gate	#41	ready, CI green, replay gate PASSED
#25 line-item review UI	#42	ready, CI green, both E2E flows green

Open decisions: #23 OCR model source (off by default); #24 gate threshold (5 points) and the gap that recorded answers cannot catch prompt regressions. Live Vertex calls are unverified (no credentials by design). Full report, merge order and follow-ups: #2 (morning report comment).

---

Generated by Claude Code

[Fluory]

Closed on the orchestrator's instruction: every issue of this epic (#22–#25) is merged into main (PRs #39–#42). Open follow-ups: #46 (line items to the ERP), #49 (OCR evals in CI), #50 (verifier across table cell borders).

---

Generated by Claude Code