Full check found the premise wrong on every count (like #71/#70):
- Not 140 docs/17,700 pages/2hr/$$ needing Dafna+chaim. Of 140 image-less
docs, only 65 are PDF (rest MD/DOCX — pipeline renders PDF only) = 704 pages.
- The value docs (appraisal, where multimodal's table/image worth is) were
already 8/12 embedded. The only gap was ONE case, 8070-25 (4 appraisal docs).
- Backfilled 8070-25 locally (voyage-multimodal-3, ~30s, cents): all 14 docs
embedded. Appraisal coverage now 12/12 (100%).
- Remaining 51 PDFs/649 pages are all text-dense (reference/response/appeal);
#15 proved multimodal does NOT help text-dense docs, so they're intentionally
left text-only. Not an inconsistency — the correct config.
No gold-set / Dafna labeling / chaim cost approval needed — cost was cents and
value was already proven in #15. #80 done (technical, not human-gated).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The 4 'ambiguous' citation items flagged for chair turned out to be dead
orphan stubs: 0 inbound/outbound edges across all 5 citation mechanisms,
0 full_text, 0 halachot, 0 chunks/embeddings. A corpus-wide check found 15
such orphans total (incl. clean-looking ones). Per OpenCitations (keep an
id-less entity only if it is CITED — these are cited by nothing), these are
pure noise → deleted, not chair-judgment.
- 15 orphan cited_only stubs deleted (cited_only 46 -> 31); backup in
data/audit/fu2b-orphan-stub-cleanup-*.json.
- 0 malformed / 0 orphans remain; all 31 remaining stubs are cited.
- Combines with the 3 earlier mechanical normalizations. #70 fully done.
- Known forward-edge (no current data, no task): '+' combined-citation
handling in citation_extractor if it recurs in future extraction.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A/B eval (eval_retrieval.py, 86-query gold-set) showed the 0.5 default was
mis-tuned: the image side was too heavy and dragged precedent_library recall
0.971 -> 0.885. Sweep 0.5..0.75 — at 0.65 multimodal beats text-only on every
overall metric AND every corpus (R@5 0.994 vs 0.989, nDCG@5 0.960 vs 0.944,
MRR 0.954 vs 0.936). Dafna approved.
- MULTIMODAL_TEXT_WEIGHT=0.65 set in Coolify (legal-ai, runtime) + redeploy.
- baseline.json updated to the 0.65 config (future regression reference).
- #15 done (premise was stale — multimodal already default on 110 docs; the
win was tuning the weight, not the backfill).
- #80 opened: the costly 140-doc legacy backfill is deferred until a targeted
image-answer gold-set proves the table/image value prop (untested here).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Discovered closing #57: the current chunker still emits 4 tiny chunks that
are standalone section headings ('דיון', 'טענות המשיבים', ...). Low priority
— filtered at query time, search unaffected. Proposed fix: anchor a short
isolated heading forward into the following section.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
#78 (committee-upload wakeup) + #77 (case_number identity) shipped.
#76 (Paperclip create-task button): root-caused to ee=companyId guard —
button enabled on title only but submit requires a company; not safely
patchable via injection. Deferred with workaround + upstream note.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Diagnosis from the FU-5 eval: co-relevant precedents for broad legal questions
rank 15-16 (retrieved, not absent — recall ~1.0 by rank 20). Tracked as a
deliberate, harness-measured tuning task rather than an unmeasured global limit
change (which affects UI + writer agents + token cost).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
חיווט ספ-המערכת לסוכני-Paperclip כך שכל סוכן חייב לקרוא את 00-constitution
תחילה, ואז את ספ-התחום הרלוונטי לתפקידו (לפי טבלת X4 §2) — לפני עבודה מהותית.
- HEARTBEAT.md: סעיף עליון "קריאת-ספ — קודם החוקה (00), אז ספ-התחום" לפני §0–§8,
עם טבלת תפקיד→ספ ל-8 הסוכנים.
- 8 קבצי-סוכן (ceo/proofreader/researcher/analyst/writer/qa/exporter/hermes):
סעיף "קרא לפני פעולה (INV-AG1)" בראש הגוף.
- X4-agents.md: שדה "אכיפה" של INV-AG1 → "מחוּוט (פרוצדורלי)"; §5 → "בוצע".
אכיפה פרוצדורלית בכוונה — invariant פרויקטלי-תפעולי, אין שער-קוד שמכריח קריאה.
prereq לסוכני-התהליך (תת-פרויקט 5). gap-audit נשמר כ-snapshot (כמו FU-8a).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Covers GAP-11 (INV-RET4/G8) and GAP-14 (INV-QA1/G10). Retrieval quality was
never measured (only telemetry observation) and the halacha review backlog was
invisible (the 10/19 gap was found by accident).
Unit B — backlog visibility (pure code, container):
- metrics.halacha_backlog(conn) → {pending_review, approved, rejected, published,
total, oldest_pending_at}; surfaced in metrics.get_dashboard() (get_metrics MCP
tool) and /api/system/diagnostics. Live count revealed 178 pending / 1552 total,
oldest from 2026-05-03 — previously invisible.
Unit A — retrieval eval harness (host-side scripts):
- scripts/eval_gold_bootstrap.py — seeds data/eval/gold-set.jsonl. Two sources:
citations (cited==relevant via search_relevance_feedback — empty until decisions
cite precedents) and known_item (query=case_name → relevant=self; a real
citation-free signal, the methodology #52 checked by hand). Idempotent; preserves
source='chair' rows.
- scripts/eval_retrieval.py — runs the production retrieval path (search_library /
search_internal) over the gold-set; computes precision@k, recall@k, MRR, nDCG@k
(k=5,10); aggregates overall + per-corpus + per-practice_area; writes a report and
a delta vs committed baseline.json (which records the retrieval_config it reflects).
--self-test unit-checks the metric math offline.
Gold-set strategy = hybrid (chair decision): bootstrap + chair review. The citation
source is empty today (0 cited precedents in decisions), so the seed is known-item
(77 queries: 54 internal_decisions + 23 precedent_library). The gold-set is
PROVISIONAL until Dafna reviews it (the domain chair-gate).
Baseline (production config: multimodal+rerank on): R@10=0.987, MRR=0.837,
nDCG@10=0.872. Finding: MULTIMODAL_ENABLED=true slightly lowers known-item recall
(image-page results displace exact name matches) — relevant to #15. precedent_library
weaker than internal (R@10 0.957 vs 1.0) — one external precedent unfindable by name.
"CI gate" realized as discipline (re-runnable harness + committed baseline + run
before/after any retrieval-layer change) — retrieval needs prod DB + Voyage, no CI
runner has that access.
Spec: docs/superpowers/specs/2026-05-31-fu5-eval-harness-design.md
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Deterministic migration of ~52 internal_committee rows whose case_number holds
a full citation → normalized bare number (citation_formatted already correct).
DB analysis (2026-05-31): clean 1-token extraction, 0 key-collisions, 0
citation↔case_number mismatches, no month-padding dups. Chair-gated reversible
migration (backup→dry-run→approve→apply). One edge for chair: 8047/23 ערר vs בל"מ.
External (#68/FU-2c) split out — its citation_formatted is inconsistent.
Verified all 11 case_law FKs use id(UUID), not case_number → rename is FK-safe.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
content_hash/indexed_hash change detection + reindex_case_law from stored
full_text (no re-OCR) + drift health-check. Verified vs 3+ sources (content-
hash change detection, RAG re-embed-on-edit). #61.2 multimodal backfill closed:
42 rows are text-ingested (document_id NULL, no source PDF) — page-images
impossible without a PDF to render.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Granularity (epic-per-fix-unit + subtask-per-gap) and dependency-aware/WSJF
prioritization both backed by ≥3 authoritative sources (SAFe/Pichler/OWASP/CVSS;
Wake-INVEST/Cohn/Agile-Alliance/Atlassian/SAFe).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Reported: an agent claimed the case had no documents because document_list
returned empty — but the documents exist. Root cause: get_case_by_number did
an exact `WHERE case_number = $1`, so any formatting variant of the number
silently failed to resolve. Verified on 8137-24 (9 docs): "8137/24",
"ערר 8137-24", leading/trailing space, and "בל\"מ 8126/03/25" all returned
"תיק לא נמצא", which the agent read as "no documents" and went blind.
Add _normalize_case_number (strip leading proceeding-type prefix to the first
digit, trim, unify '/'→'-') and a normalized fallback in the lookup query
(exact match preferred via ORDER BY). One fix covers every case_number-scoped
tool (document_list, extract_references, search_case_documents, get_claims,
drafting, ...). Bogus numbers still correctly resolve to "not found". (#58)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Root cause of "agent can't find the Agasi decision in the corpus" (CMPA-55):
the decision was fully ingested, but the retrieval layer failed on the
realistic agent query — searching by case name.
- RC-A (#52): lexical tsvector covered only chunk content + halacha text,
so a bare-name query ("אגסי") matched decisions that *cite* the case, not
the case itself. Add meta_tsv on case_law(case_name, case_number) (SCHEMA
V20) and OR it into the lexical halacha/chunk SQL with a match boost, so a
name/number hit surfaces the case's own rows. Agasi: rank 4 → rank 1.
- RC-B (#53): precedent_library_list hard-defaulted source_kind=external_upload
and never exposed the param, hiding uploaded ערר/בל"מ (internal_committee)
decisions. Thread source_kind through service → tool → MCP tool (supports
'internal_committee' / 'all_committees').
- #54: agent instructions (researcher/analyst/writer) — search-by-name
protocol: add content/case-number, search both corpora, use all_committees
before declaring "not in corpus".
- #55: chunker produced tiny fragment chunks ("דיון", "החלטה") from header
keywords matched mid-sentence. Anchor SECTION_PATTERNS to line start +
merge sub-min sections; exclude <50-char fragments at query time (484
existing fragments hidden; full re-chunk tracked as #57).
Tests: scripts/test_retrieval_by_name.py (name ranks case above citer +
substantive regressions); chunker unit checks (0 tiny chunks). New findings
filed as tasks #56 (halacha source_kind leak) and #57 (re-chunk migration).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Brings the legal-ai ↔ Paperclip integration in line with the official
Paperclip skill. Net effect: HEARTBEAT.md -47% (370→195 lines), all 14
agents on uniform runtime_config + budget + instructionsBundleMode, and
two cross-company helpers replacing manual SQL.
Highlights:
- HEARTBEAT.md refactor: project-specific only, delegates to the official
paperclipai/paperclip skill (loaded per agent). Adds heartbeat-context
fast-path (§1.7) and PAPERCLIP_WAKE_PAYLOAD_JSON shortcut (§1.5).
- Issue Thread Interactions API: legal-ceo.md now uses
ask_user_questions / request_confirmation / suggest_tasks instead of
free-text comments — gives chair structured UI with idempotency keys.
- pc.sh + paperclip_api.pc_request: every API call goes through helpers
that inject Authorization + X-Paperclip-Run-Id (audit trail).
- sync_agents_across_companies.py: master(CMP)→mirror(CMPA) sync via
Paperclip API, idempotent, with --verify and --apply modes.
- skills/new-company-setup: 11-step blueprint distilling all 11 gaps
into a single onboarding runbook for the next company.
- .taskmaster: 12 tasks covering each gap (one already closed: #29).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three improvements to the precedent library based on usage feedback:
1. Auto-fill metadata at upload time. New service
precedent_metadata_extractor reads the ruling's full_text and
suggests case_name (short), summary, headnote, key_quote,
subject_tags, appeal_subtype. The merge policy fills only empty
fields, preserving everything the chair typed in the upload form.
Wired into the ingest pipeline; also exposed as a re-run endpoint
POST /api/precedent-library/{id}/extract-metadata for existing
records.
2. Edit sheet in the UI. Pencil icon on each library row opens a
pre-populated form covering every field. A Sparkles button on the
sheet runs the metadata extractor on demand and refreshes the
form. The case_number is read-only because halachot are FK'd to
it; renaming requires delete + re-upload.
3. Halacha extractor branches on is_binding. Sources marked binding
(Supreme/Administrative) keep the strict halacha prompt. Non-binding
sources (other appeals committees, district courts on planning
matters) get a different prompt that extracts applications,
interpretive principles, and persuasive conclusions — labeled with
new rule_types 'application' and 'persuasive'. The fallback also
widens chunk selection: if the chunker labeled nothing as
legal_analysis/ruling/conclusion, we now run on all chunks rather
than returning zero halachot for a usable ruling.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a third corpus of legal authority distinct from style_corpus
(Daphna's prior decisions for voice) and case_precedents (chair-attached
quotes per case). The new corpus holds chair-uploaded court rulings and
other appeals committee decisions, with binding rules (הלכות) extracted
automatically and queued for chair approval.
Pipeline (web/app.py + services/precedent_library.py):
file → extract → chunk → Voyage embed → halacha_extractor → store +
publish progress over the existing Redis SSE channel.
Schema V7 (services/db.py): extends case_law with source_kind +
extraction status fields under a CHECK constraint pinning practice_area
to the three appeals committee domains (rishuy_uvniya, betterment_levy,
compensation_197). New precedent_chunks (vector(1024)) and halachot
tables (vector(1024) over rule_statement, IVFFlat indexes, gin on
practice_areas/subject_tags). Halachot start as pending_review; only
approved/published rows are visible to search_precedent_library.
Agents: legal-writer, legal-researcher, legal-analyst, legal-ceo,
legal-qa get search_precedent_library. legal-writer prompt explains
the three-corpus distinction and CREAC use; legal-qa now verifies that
every cited halacha resolves to an approved row in the corpus.
UI: /precedents page with four tabs — library / semantic search /
pending review (J/K nav, A/R/E shortcuts, badge count) / stats.
Reuses the existing upload-sheet progress + SSE pattern.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- StatusBadge: added icons (lucide-react) and Hebrew descriptions for all 13 statuses
- WorkflowTimeline: added phase icons and current-status description display
- StatusGuide: new collapsible component showing all statuses grouped by phase with explanations
- StatusChanger: new dropdown for manual status override on the case detail sidebar
- Case detail page: merged action buttons into overview tab, removed separate actions tab
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Drop the defaultOpen={i===0} on the first threshold_claim — when a
case has a lot of material already on screen (research background
+ chair positions + now precedents), auto-opening the first card
creates a wall of text on page load. All cards now start collapsed,
same as the issues list already did.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Surface the new POST/GET/DELETE /api/cases/{n}/precedents endpoints
in the compose screen as two insertion points:
1. A new case-level card "פסיקה כללית לדיון" at the top of the
main column, for precedents that support the discussion intro
rather than a specific threshold_claim / issue.
2. An inline "פסיקה תומכת" section inside each SubsectionCard,
below the ChairEditor.
Both insertion points render a `<PrecedentsSection>` which shows a
list of `<PrecedentCard>` (citation + blockquote + optional chair
note + 📄 chip if a PDF was archived) followed by a `<PrecedentAttacher>`
popover trigger.
The Attacher is a Popover with cross-case typeahead: typing 2+
characters into the citation field hits /api/precedents/search and
shows distinct library matches; picking one prefills quote + chair
note but leaves them editable so customizing the quote for this
case doesn't mutate the library. An optional PDF/DOCX/DOC file can
be attached — it uploads first via POST .../upload-pdf and the
returned document_id is passed into the precedent create call.
The parent compose page issues a single useCasePrecedents query
and partitions the result by section_id into a Map so each
SubsectionCard renders its own slice without re-fetching.
shadcn Popover installed as a new primitive. sonner toasts wired
for success/error in both attach and delete flows.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Three user-reported bugs on /cases/[caseNumber]:
1. Documents tab showed "9 מסמכים" in the count but rendered nothing —
DocumentsPanel was reading filename/category/status/size_bytes/uploaded_at,
but the real FastAPI payload (case_get → db.list_documents) returns
title/doc_type/extraction_status/page_count/created_at. Rewrote the
panel against the actual document row shape, added a CaseDocument
type alias in lib/api/cases.ts, mapped doc_type to Hebrew labels
(כתב ערר / כתב תשובה / ...) and extraction_status likewise.
2. The "פעולות" tab showed a debug-flavoured paragraph "עריכת פרטי התיק
נשמרת מיד דרך PUT /api/cases/1033-25" — that was internal wording,
not user copy. Removed.
3. Overview tab showed the raw enum value "full_acceptance" in the
expected-outcome line. Mapped through the existing expectedOutcomes
label array so it now reads "קבלה מלאה".
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Close out the read-only surface before cutover with three families of
small fixes that the previous phases left unfinished:
- Error boundaries: add src/app/error.tsx (route-segment), global-error.tsx
(root crash fallback with its own minimal html/body — no Providers
dependency since those may be the thing that crashed), and not-found.tsx
for a Hebrew 404 instead of the default Next page.
- Accessibility: wire usePathname() into AppShell so the current nav item
gets aria-current="page" and a gold underline. Add aria-label + aria-hidden
on the icon-only buttons that Phase 5 left text-less (corpus trash,
parties-field Plus). Nav gets an aria-label of its own.
- Metadata template: title on each route now reads "X · עוזר משפטי" via
the layout.tsx title.template. Description localized to Jerusalem.
- README: full E2E smoke test checklist covering all 9 screens, plus a
backend contract table so future phases know which hook wraps which
endpoint. Documents the known Gitea→Coolify webhook issue.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Port the remaining read views from the vanilla UI to Next.js:
- /diagnostics — system health snapshot (DB connected, table counts,
active tasks, failed and stuck documents). Uses the existing
/api/system/diagnostics payload with a 10s refetchInterval so the
page self-updates while the user watches.
- /skills — Paperclip skill inventory with sync status (DB-only,
disk-only, synced, not-synced) as a card grid driven by
/api/admin/skills.
- /training — Dafna's style portrait as three tabs on one page:
* Report: corpus KPIs + CSS conic-gradient subject donut
(SubjectDonut ported from index.html renderHero) + horizontal
anatomy bars + top-12 signature phrases.
* Corpus: TanStack Table of style_corpus rows with an inline
delete mutation (useDeleteCorpusEntry invalidates both the
corpus list and the style report so KPIs update).
* Compare: two-decision selector backed by /api/training/compare,
side-by-side panels plus shared / only-A / only-B pattern
lists.
New API modules: lib/api/system.ts, lib/api/skills.ts,
lib/api/training.ts. All three use TanStack Query with staleTime
profiles tuned per endpoint (10s for diagnostics, 30s for skills,
60s for training reports).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Backend commit 26d09d6 introduced a multi-tenant axis (practice_area +
appeal_subtype) that the vanilla UI picked up but the new Next.js
rewrite did not. Close the gap in the screens we already shipped so
future search/filter work in Phase 5 has the right data on screen.
- lib/practice-area.ts — new: enum + label maps + deriveSubtype(),
mirrors mcp-server/src/legal_mcp/services/practice_area.py.
- lib/schemas/case.ts — two new z.enum fields on caseCreateSchema.
- lib/api/cases.ts — Case / CaseDetail gain practice_area and
appeal_subtype as optional (cached pre-migration responses still
parse cleanly).
- wizard/case-wizard.tsx — basics step gains a practice_area dropdown
(future domains disabled with "(בקרוב)") and an appeal_subtype
dropdown with auto-fill effect tracking a userTouchedSubtype ref,
same behaviour as wireSubtypeAutofill() in the vanilla UI.
- cases/case-header.tsx — gold badge next to the status shows
"ועדת ערר · רישוי ובנייה" when both fields are populated.
- cases/cases-table.tsx — new "תחום" column showing subtype label
(dash for unknown). No filter yet — that's phase 5 when a second
domain actually exists.
Plan: ~/.claude/plans/woolly-cooking-graham.md
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The committee_type field in FastAPI's CaseCreateRequest is a leftover
with no meaningful UI. Appeals against a ועדה מקומית / ועדה מחוזית are
legally distinguishable by case-number range, not by a picked committee
label; appeals against a ועדת ערר decision go to בית משפט לעניינים
מנהליים and never enter this system at all. The backend retains its
"ועדה מקומית" default, which is what we'd send anyway.
Drop the Select from the wizard's basics step and the "ועדה" row from
the case detail header. The Case TS type keeps the optional field so
existing API responses still parse cleanly.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add dialog/select/textarea/label/progress/sonner components and wire
a Toaster into Providers. New zod schemas in lib/schemas/case.ts
mirror CaseCreateRequest/CaseUpdateRequest and feed react-hook-form
validation.
CaseEditDialog on the case detail Actions tab posts PUT /api/cases/{n}
with optimistic cache patching via useUpdateCase, showing toast
feedback on success/error.
shadcn's <Form> registry entry skipped at init (missing from the
nova preset); the dialog uses RHF directly against the same Input/
Textarea/Select primitives.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Initialize shadcn/ui (radix-nova preset) and wire its semantic tokens
to the editorial navy/cream/gold palette so primitives inherit the
judicial voice without per-component overrides.
Replace the Phase 2 live-probe with a real dashboard: KPI tiles,
conic-gradient status donut (ported from the vanilla renderHero),
and a TanStack Table cases list with search + sort. Add useCase(n)
hook with 5s staleTime/refetchInterval to replace the old manual
polling loop when Case Detail ships next.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- create-next-app with TypeScript, Tailwind v4, App Router
- Port design-system.css tokens into Tailwind @theme (navy/gold/parchment, Heebo)
- Install TanStack Query, react-hook-form, zod, lucide-react, react-dropzone
- layout.tsx: RTL Hebrew + Heebo via next/font/google
- AppShell component with navy header + gold rule + nav
- next.config.ts: output:standalone + rewrites to proxy /api/* to production FastAPI
- Dockerfile: multi-stage Node 20 Alpine build for Next.js standalone
(branch-local override of the FastAPI Dockerfile; main is unaffected)
- Switch .taskmaster to claude-code provider (no API key required)
- Add 7 phase tasks (83-89) tracking the full rewrite plan
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
