diff --git a/docs/product-technical-gap-baseline.md b/docs/product-technical-gap-baseline.md
index e65883463..c49404ae5 100644
--- a/docs/product-technical-gap-baseline.md
+++ b/docs/product-technical-gap-baseline.md
@@ -1,26 +1,266 @@
# Product & Technical Gap Baseline
-## 1. Known Parsing & Frontend Display Gaps
-- **Footnote Parsing**: `post=00505695-3e61-1fd1-83c5-263f88a9e77a` fails to recognize footnotes (li/oi level errors).
-- **Table Parsing**: `post=00505695-3e61-1fd1-80c6-86bb61c8ddc5` completely fails at parsing tables.
-- **Indentation**: Incorrect indentation rendering in `post=00505695-7571-1fd1-83c3-d521b187ad5b` and `post=00505695-3e61-1fd1-83c0-497b3c1c455e`.
-- **Image/Table OCR**: `post=00505695-7571-1fd1-83dd-3d22a61a5734` fails text recognition for tables inside images, markdown parsing fails, and image OCR description is too shallow for Ontology & Semantics.
-- **Math/Superscripts**: `post=00505695-9612-1fe1-83a7-e30153323f25` fails to parse superscripts like m^3 properly. Needs strict Ontology grammar for math formulas.
-- **Missing UI Elements**: DAG (Directed Acyclic Graph) view is currently missing from the frontend for `post=00505695-7571-1fd1-83c5-895ed333cdbc`.
-
-## 2. LLM Extraction & Knowledge Graph Gaps
-- **Multiple Project Extraction**: (Resolved) LLM prompt updated to request key_events as objects with project_name, separating events correctly.
-- **5W1H Missing**: (Resolved) LLM prompt updated to explicitly request 5W1H evidence items in the JSON output array.
-- **R&R and Keyman Missing**: (Resolved) LLM prompt updated to explicitly instruct using actual stated names rather than collective titles.
-- **Entity Resolution / Searxng**: Abbreviations like "한전" and "한국전력" are not mapped properly using Searxng and KG corroboration.
-- **Meso-level Team Mapping**: (Resolved) Checked extraction logic; `team` mapping logic is present and correct, but LLM needed better explicit instruction which is covered by R&R resolution.
-- **Base64 Image Omni-modal**: Current text-only embedding fails on images. Omni-modal LLM processing is required for images to capture layout, font size, colors, and spatial meaning.
-
-## 3. General Architecture Gaps
-- **DB Architecture**: Ensure PostgreSQL is strictly used (no file DBs), 3rd normal form is maintained, and Hot Partitions are handled. DB locks must be managed (or use read/write replicas).
-- **Zotero Integration**: Papers and standards referenced by TEPP must be synced via Local Zotero API (http://localhost:23119/api/) and cited using APA 7th edition in docstrings.
-- **Testing**: We need actual testing of Psychometrics (Fast-MLSIRM parameter calibration, RMSE of estimates, Fixed-Item Parameter Calibration, CAT) against synthetic/demo data.
-- **Security & Compliance**: PII masking cannot break the system. Need SOC 2 and CSAP compliance alternatives to blind PII masking.
-- **LLM Orchestration**: Ensure ALL LLM calls route through `contextual-orchestrator` utilizing API keys (BYTEZ, NVIDIA, OPENROUTER, OPENAI) with auto model discovery and optimal reasoning effort allocation (Fugu/Conductor/TRINITY research).
-
-*This document is continuously updated by the hourly automated agent loop.*
+> Audit snapshot: 2026-08-24 03:19 KST. This repository records synthetic
+> fixtures and aggregate, non-identifying runtime evidence only. Open PRs and
+> local checks are not protected-default-branch release evidence.
+
+## 1. Exact-head and governance evidence
+
+The protected default branch was
+`ef6f5a5ffcb467bd935dc1e53acc0029669b0bd7` when this baseline was refreshed.
+The live queue contained 53 open PRs and no independently approved head.
+Twenty-two current heads had an aggregate failed check state, four had pending
+checks, and no current head retained an unresolved review thread.
+Those results must be re-fetched against the current head before remediation or
+a merge claim.
+
+Recent protected-default-branch evidence:
+
+| PR | Result | Evidence boundary |
+| ---: | --- | --- |
+| #373 | merged as `cf419cd90676a7f8cf82fac356d4ee24468017a7` | protected merge verified after the live queue refresh |
+| #420 | merged as `c7f4d6c1133e680befe076442f16ba5f1c722405` | protected merge verified; later main changes still require their own acceptance |
+| #475 | merged as `02fe67005f25f2025ce29b403c42097838b751ad` | restored a non-identifying baseline; a later broad merge overwrote it and this repair removes the regression |
+| ContextualWisdomLab/.github #1248 | merged as `9ad0ad50409561292b424d6f35a95d670a277e77` | central Strix scope repair is available to subsequent reruns |
+
+The organization scheduler is the single review/repair control plane. Its
+quarter-hour queue sweep and hourly heartbeat satisfy the hourly loop
+requirement without a duplicate repository-local scheduler.
+
+## 2. User-visible capability baseline
+
+Substantially present in source or active PRs:
+
+- PostgreSQL-backed import, normalized provenance, cutoff-aware analysis runs,
+ source revisions, lineage reconstruction, and explicit unavailable states.
+- Authenticated workspace navigation, post detail, localized summaries, 5W1H,
+ R&R/Keyman, evidence citations, chat, organization hierarchy, and lineage DAG.
+- Semantic paragraph/list/table/image-region units that preserve source and
+ provenance instead of flattening a record into one body string.
+- Contextual-orchestrator boundaries for adjudication, extraction, summaries,
+ chat, embeddings, and VISION; null channels remain unavailable and are
+ dropped from score fusion.
+- W3C PROV-O projection through normalized provenance tables, with the
+ knowledge graph retained as an explicit navigation projection.
+
+These statements describe source capability, not authenticated production
+corpus acceptance or protected release.
+
+PR #486 merged as `f11a2cb546792622932011587fe6f6aa54c79948`
+into non-default `docs/customer-master-scope-adr`. Its documentation follow-up
+is branch commit `5ec42616e6b4ab6ed0b7757c299523110acb62ca`, but no open PR carries that
+branch to `main`; both remain stack evidence, not protected-main release
+evidence.
+
+Subsequent stack evidence, verified at 2026-08-23 22:44 KST, also remains
+outside protected `main`: PR #491 at
+`1a5139f233f2dc484f970b25c4e1b51462dbedf8` merged normally as
+`60efee9b9a97fa433f1d8a83f396c7f8d55df39e` into the non-default
+`worktree-fix-frontend-build-break` branch. Its four-file documentation diff
+records ADR 0157's public ontology namespace identity without a production
+namespace rewrite. The exact stacked head passed the hosted full and frontend
+suites, but the target-repository central review workflows did not run for
+that non-default target. This is stack integration evidence, not evidence of a
+protected `main` release or independent approval.
+
+At 2026-08-23 23:00 KST, PR #492 at
+`38f3734f58477bad04f82758fc685e798e9d4b7b` also merged normally as
+`c8a4be8fc2417f05d53fb68d32d9e59c3d443e25` into that same non-default
+branch. Its six-file, +166/-2 compatibility-publication slice is intentionally
+minimal, but remains stack evidence only; it is not a protected-`main` merge
+or independent approval.
+
+The official repository Pages API was activated at 2026-08-24 00:53 KST with
+`build_type=workflow`, public visibility, HTTPS enforcement, and repository-case
+`html_url` `https://contextualwisdomlab.github.io/LineageWeave/`. Live
+publication nevertheless remained incomplete: the lowercase
+ontology URL, repository-case ontology URL, and repository-case
+`namespace-compatibility.ttl` URL each returned HTTP 404. Source and active-PR
+publication contracts therefore do not prove an available public endpoint.
+Cross-repository organization-site PR #188 at
+`4037d5fe6176f8cb78aa41696bc3dc50ff8a7f32` provides the owned lowercase
+route regenerated from LineageWeave source head `9b47ce42`. Four route tests,
+RDF isomorphism, exact manifest inventory/hash/source commit, and upstream byte
+equality passed. The PR remains open with hosted Strix and OpenCode checks
+pending, no unresolved thread, and no exact-head approval, so that result is
+not protected deployment evidence and the live HTTP 404 remains authoritative.
+
+The central Strix dependency is now superseding ContextualWisdomLab/.github
+PR #1263 at `9ca02e84e1625a674aad4bdcb392d09b641c88aa`. The reviewed source
+normalizes the direct-OpenAI fallback at the Strix/LiteLLM boundary and includes
+the NVIDIA fallback doctoring path in changed-path validation. Twenty-six
+focused pytest cases, the full C-locale shell harness, actionlint, and shell
+syntax passed before the latest concurrent test-only head advance. All reviewed
+threads were resolved, but hosted exact-head checks remain pending and it has
+no approval. Therefore older failed runs that executed trusted `main`, and
+local green evidence, do not authorize transfer of a Strix result.
+
+Current active-PR evidence remains outside protected `main`:
+
+- PR #258 at `f0b5234db6d3b8ef76cfd98126f6d7165e9308dc` carries ADR 0146's
+ distributed MCP principal rate-limit decision. Its current MCP lifespan uses
+ the standard-library `AsyncExitStack` so a limiter-factory failure closes an
+ already-open database pool and normal shutdown remains ordered. Forty-seven
+ focused MCP tests passed. It had no unresolved thread or exact-head approval;
+ the current Devin failure supplied no review, comment, or annotation.
+- PR #426 at `c4a16c20bbdd50cbf51d596448e4c09fe38cea31` is the current
+ prerequisite carrier for ontology publication and the merged ADR 0157 and
+ compatibility-publication stacks. ADR 0036 now explicitly names the OWL 2
+ Full/RDF-Based semantics required by the retained `rdf:Statement` contract,
+ and the PROV support-profile IRI is canonical lowercase. Seventy-seven
+ focused ontology tests, deterministic publication, and profile byte-equality
+ passed on source head `9b47ce42`; later stack merges add the reviewed two-axis
+ leftover map, observed/expected residual operands, and optional-extra
+ collection repair. It had no unresolved review thread or exact-head approval;
+ hosted checks were pending. Results from prior heads are stale, and Pages
+ activation alone does not deploy this unmerged source.
+- PR #355 at `6fc22a9471bfb4d94b18f884e012cd823b296382` carries ADR 0145's
+ Naruon calendar projection boundary. Its aggregate checks failed, with no
+ unresolved thread or exact-head approval.
+- PR #417 at `c5c0929c68abf876de6c924d5a93b554d6bcadfb` preserves stale image
+ summaries across provider refresh failures. Two focused unit tests and two
+ API tests passed; it had no unresolved thread or exact-head approval.
+- Stacked PR #494 at `5d9728a16051e7db453ca513cd5baa75be7450cc`
+ merged normally as `1ff0cd13b84d5c5f817706ef23dcbd5c3d67a510`
+ into non-default #426. It keeps optional-extra collection fail-closed,
+ tolerates non-UTF-8 optional source, and restores the shared OIDC login gate.
+ Ten focused tests and repository collection of 430 tests passed on the merge
+ SHA. This is stack evidence, not protected-main evidence.
+- PR #493 at `499c8b1bc4cdffff5bb985658bcb3821d312cedb` includes product-fix
+ head `612c4cc1`, where the focused backend suite passed 13 tests, the App
+ suite passed 81 tests, and lint plus diff-check were clean. The later head
+ had no unresolved thread or exact-head approval, while hosted checks
+ remained pending.
+- PR #496 at `78287c08309f614ca1de04612c3e15c555bed1c6` carries savepoint,
+ normalized TEPP analysis-run identifier, batched list/detail receipt, state
+ progression, and replay-contract fixes. The current head also isolates the
+ older optional outbox and reconstruction reads in savepoints so a missing
+ optional table cannot poison the caller transaction. Its latest change keeps
+ a durable accepted receipt Running across an unpersistable re-check; the
+ exact-head receipt suite passed 21 tests.
+ It had no unresolved thread or exact-head approval, while hosted checks were
+ pending; local results are not protected-release acceptance.
+- PR #498 at `35823d889c5360ebf2152ed5679d7c22d6832545` retained the public
+ health probe, login return, and production docstring gate. Its informational
+ review threads were resolved; it had no exact-head approval and hosted checks
+ remained pending. This is source and test evidence, not a live-runtime claim.
+- PR #439 at `5db6276c53e8f59499dbac0f59f6cbf0555b99f2` fixes the shared
+ lineage-layout walk so a visible root feeding a cycle terminates, with a
+ synthetic rooted-cycle regression. Ninety-five focused frontend tests and
+ TypeScript compilation passed before the guarded fast-forward push. Its later
+ informational thread was resolved; exact-head approval was absent, and hosted
+ checks were pending.
+- PR #501 at `b895aa57d8b7c8b70f8830ecf32a44972199e6a1` preserves the lineage
+ channel-weight fallback transaction with the same savepoint boundary at both
+ callers. Its informational thread was resolved; it had no approval and hosted
+ checks were pending.
+- PR #435 at `36164fbcc1bb15b8fab991dbe5629f719b74896c`, PR #500 at
+ `487b5b487dddf2d40df71f6add88c7f07347c086`, and PR #502 at
+ `accc33f1343a2ae3e5ad04447e7fce861c82d9b9` now share the existing safe
+ OIDC return-URL persistence and remove the unreachable unauthenticated Admin
+ render. Their exact composed heads passed frontend lint, respectively 140,
+ 140, and 147 tests, and production builds. Hosted checks restarted; none had
+ an exact-head approval. Their aggregate hosted failures still require
+ exact-head diagnosis before release.
+- Stacked PR #504 at `166436d519341aa10ec9b32b4d67b1a24818f5c8`
+ merged normally into non-default #426 and stops the PROV-O support profile
+ from minting repository-case product IRIs. Its 67 focused tests and
+ deterministic site build passed, but the stack merge is not protected-main
+ release evidence. The stored project-mention migration remains deferred
+ until #428, #484, and #426 reach protected `main`.
+- PR #428 at `f81f831c14aa080f671fbdb781bf8afecdd36e82`
+ makes migration 0103 replay idempotent while preserving operator-customized
+ synthetic catalog rows. Five focused static tests and a twice-applied local
+ PostgreSQL replay passed; hosted evidence remained pending and it had no
+ exact-head approval.
+- PR #507 at `b8d9ce429f223d43a8639d0e2b2b0777e9105d2b`
+ reuses that migration replay fix and the existing safe OIDC return helpers.
+ Its review follow-ups reused existing imports and a named helper; the latest
+ focused suite passed 13 tests. It had no unresolved thread, failure, or
+ exact-head approval; hosted checks restarted.
+- PR #490 at `63f3231d249f20b0f34f7fc56fbd3f28f62f6d0c` is the current carrier
+ of the consolidated product stack and includes ADR 0143. ADR 0144 was absent
+ from both this exact tree and protected `main`, so it is not attributed to
+ this head. The validated post-merge #505 migration correction was transplanted
+ here because #505 merged before that source-branch commit; fifteen migration
+ replay tests passed. Hosted checks restarted, with no unresolved thread or
+ exact-head approval.
+- Stacked PR #509 at `bba8a8ac43a43db70c563dd9612ab74c3fbe7930`
+ was normally restacked and merged as `e4d692c6e5daede2af7c0e259d3fc5a4c1c7636a`
+ into non-default #490 after its duplicate migration slice conflicted.
+ Ninety-six focused tests and twelve PostgreSQL schema tests passed; that stack
+ merge is not protected-main evidence.
+- Stacked PR #512 at `0e8f27151b25a7b4f221832b8bdb285e2855e1ab`
+ merged normally as `2ff02bb50280978c3720639d16c10494b4e0eeb7`
+ into non-default #490 after terminal green tests and zero unresolved threads.
+ It requires project evidence before admitting a planned-facility relation;
+ this is stack evidence, not protected-main evidence.
+- Stacked PR #508 at `75565bf96d12c4e8f49dcb4ce61b9a0f24e2c929`
+ merged normally as `f6f295057eea649caf00fa119345c43cb57c2415`
+ into non-default #426. It reuses the shared OIDC gate and measures leftover
+ distance on the first two retained Gabriel axes; 140 frontend tests, lint,
+ typecheck, and build passed. This is not protected-main evidence.
+- Stacked PR #511 at `9227477774565ea2a89b52c38769c0080060e0c7`
+ merged normally as `4dd8b364bb2ea2db036e47c15dd87eff2b560d42`
+ into non-default #426. It aligns the Python reconciliation boundary with the
+ database's strict `1e-6` constraint and persists observed/expected leftover
+ operands without fabricating historical values. PostgreSQL, migration, API,
+ frontend, lint, and build checks passed; this is not protected-main evidence.
+- Stacked PR #513 at `2b9dfacfd66707b7920ffc5b02efdbbde008a68a`
+ keeps migration 0164 as the sole sequential source for leftover-map rank and
+ composes it with #508's two-axis distance and #511's observed/expected
+ operands. The composed head passed 144 backend tests, 179 frontend tests,
+ lint, build, and 37 latest-base focused tests. Its hosted full, frontend,
+ CodeRabbit, and Devin checks then reached terminal success, and it merged
+ normally as `c4a16c20bbdd50cbf51d596448e4c09fe38cea31` into non-default
+ #426. This remains stack evidence, not protected-main evidence.
+- Stacked PR #514 at `b90d976270036adce8464434836d7e1541ba5223`
+ merged normally as `d02ab8fda2a6d561991e0f88f71e18d5dafe4a49`
+ into non-default #490 after documentation hygiene passed. Its effective diff
+ removed one trailing ADR blank line; it is not protected-main evidence.
+- Stacked PR #515 at `2d11b4b87beed3eaa7e452349a2daedb44cc32f7`
+ carries quantity superscript display work onto non-default
+ `feat/quantity-superscript-display-v2129`. Fifty-one backend tests, forty
+ frontend tests, lint, build, Ruff, and diff-check passed; hosted full and
+ frontend suites also succeeded. Devin remained pending, so the exact head
+ was not terminal green and was not merged.
+
+## 3. Open product and technical gaps
+
+| Gap | Current evidence | Acceptance requirement |
+| --- | --- | --- |
+| Protected release | 53 PRs remained open, none had an independent current-head approval, twenty-two current heads had an aggregate failed check state, four had pending checks, and none retained an unresolved review thread | Terminal exact-head checks, no unresolved threads, independent approval, and a protected merge SHA |
+| Authorized-corpus runtime | Repository tests use synthetic fixtures; private records remain outside git | Authenticated runtime validation returning only aggregate, non-identifying evidence |
+| Image understanding | Region, OCR, and description work exists across active heads | Orchestrator-backed rendered workflow, original/derived asset provenance, and honest unsupported states |
+| Semantic source rendering | Paragraph, table, list, and formula parsing exists across active stacks | Authenticated browser evidence that semantic units render without authoring-layout artifacts |
+| Scientific measurement | TEPP and fast-mlsirm adapters are present or under review | Persisted accepted envelopes, calibration/recovery evidence, and no invented theta |
+| Accessibility and responsive UX | Unit coverage exists for major user surfaces | Keyboard, screen-reader, mobile, and authenticated Playwright acceptance on the exact release head |
+| External integrations | Search, Zotero, calendar, and downstream consumer contracts are bounded | Provider conformance, failure/reconciliation behavior, and provenance-bearing integration evidence |
+| Release quality | Local focused/full suites have passed on individual PR heads | Repository-wide coverage, docstrings, Storybook, security, browser, and release evidence on one exact head |
+
+## 4. Evidence boundaries
+
+- Never add a real record, title, name, identifier, screenshot, log, benchmark
+ artifact, or documentation example to this repository.
+- Attendance or co-occurrence is not responsibility, project, customer, or
+ affiliation evidence. Preserve uncertainty and provenance.
+- Missing transport, model capability, accepted envelope, or persistence is
+ unavailable or failed evidence, never a placeholder result.
+- Local green tests, bot statuses, auto-merge, and warning-only checks do not
+ prove a protected merge.
+- Re-fetch base/head SHAs, checks, review threads, approvals, rulesets, and the
+ merge SHA immediately before any lifecycle claim.
+
+## 5. Next acceptance loop
+
+1. Reproduce and repair the current Strix failure without weakening the
+ scanner or transferring results across heads.
+2. Re-fetch every open head, latest checks, unresolved threads, and independent
+ reviews before any merge claim.
+3. Run frontend lint/test/build/Storybook, backend tests, and authenticated
+ browser/accessibility checks on the exact candidate release head.
+4. Reproduce user cases with synthetic fixtures or authorized aggregate
+ runtime evidence, preserving `unavailable` explicitly.
+5. Fix only evidence-backed failures and repeat the protected merge gate. Do
+ not self-approve, force merge/push, bypass protection, or transfer stale
+ review/check evidence across heads.
diff --git a/frontend/src/App.tsx b/frontend/src/App.tsx
index 6fba0dd41..1b5b351ab 100644
--- a/frontend/src/App.tsx
+++ b/frontend/src/App.tsx
@@ -4610,7 +4610,8 @@ export default function App({ showLabPanels = false }: { showLabPanels?: boolean