# V1 teaching and design completion · active work · 8 September 2026

## Objective and full scope

Audit and enhance the complete V1 release line in Presentation Center. Cover
member behavior, underlying mechanisms, creator and operator workflows,
implementation qualifications, failure/recovery and acceptance. Use appropriate
source-supported diagrams and concrete cases throughout. The existing guide
count is a baseline, not a limit on the final teaching program.

The baseline is `c8b82d6fd87184503a1b0d3dcc40309f632f59cc`: 60 library guides,
578 unique slides and 348 cited paths, with current generated artifacts verified
by `build-eve-oshun.py --check`. The non-Eve V1 material occupies 28 guides, 310
guide-slide instances and 292 unique slides. The orientation guide and the V1
portfolio introduction are also in scope. Shared slides must be reviewed in
every affected guide. Preserve the completed Eve course and audit its V1
connections and source qualifications rather than creating a duplicate course.

The [audit baseline](v1-audit-baseline.json) records all 134 tracked V1 Markdown
sources and 2,929 meaningful source units at the implementation baseline. All
134 files are byte-identical to the primary documentation snapshot
`1ceae52c7896a59ac74a4429d410f3218f0b91c6`. This establishes documentation
continuity, not current code/deployment completeness. Follow the supporting
contracts, code, tests, walkthroughs and workbench inventories for each subject.
Existing citations are discovery links; they are explicitly not a reviewed
teaching-coverage claim. The existing inventory and assignments remain the
canonical unit-accounting system.

## Audit findings so far

- The complete eight-page Workbenches handbook, including the seven-initiative
  estate and Yemaya Study Workspace, has no slide citations. These are real
  teaching gaps to resolve; the current Studio guides are not a substitute.
- Older product prose describes six-domain navigation and a five-domain launch
  set. The current release contract narrows the member surface to Tara, Nyx,
  Arete and Nisaba; Veritas/Metis remain V1.2 and native/fiat remain V1.1.
  Review code and the governing brand record before repeating older summaries.
- The web/PWA pilot still has planned compositions that never reached slides:
  two distinct rooms share one column, shell ownership is shown as layers, and
  several detailed mechanisms are compressed into lists. The first delivery
  package finishes this guide and identifies the separate account/surface depth
  it needs.
- Source coverage across the other feature pages is largely file-level. A file
  appearing in citations does not demonstrate that its meaningful sections are
  taught. Every source/guide remains pending until its explanation is reviewed.

## Design and story

Visual thesis: an illustrated field guide to Lilith and the V1 platform, using
cream paper, Georgia claims, readable Arial evidence and restrained terracotta
relationships. Make member choices tangible; make ownership, state, data and
recovery explicit. Use real product imagery when recognition helps, and native
selectable diagrams for mechanisms and exact relationships.

Content plan: orient to the release and a member outcome; explain the shell and
account; deepen the four launch rooms; follow their shared evidence, memory,
runtime, creation, governance and delivery dependencies; then teach the V1.1 and
V1.2 additions with their exact release boundaries. Give workbenches, design,
operations and other large subjects their own coherent guides where the audit
shows that an overview cannot carry the material. Each guide needs a concrete
case, useful mechanisms and a recovery/verification close.

Interaction thesis: retain reduced-motion-aware slide entrance, source reveal
and workspace navigation. Add interaction only when it helps a reader inspect a
real diagram or artifact; HTML/PDF must retain the complete explanation without
interaction. Product-image panning must preserve deck navigation and keyboard
access.

## Required work and completion evidence

| Workstream                            | Required outcome                                                                                                                                        | Status                                                                                                                                                                                                             |
| ------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Source and teaching audit             | Every V1 source family and meaningful subject has reviewed teaching or an explicit, substantiated cross-reference; reconcile code/document differences. | Active; baseline recorded, substantive review underway.                                                                                                                                                            |
| Member foundation                     | Product/release/surface/account/PWA explanations and diagrams; honest offline, identity and recovery cases.                                             | Foundation, account and offline guides delivered; integrated review remains active.                                                                                                                                |
| Member rooms                          | Tara, Arete, Nyx, Nisaba, plus later Veritas and Metis: complete feature mechanisms, data, concrete journeys and failure paths.                         | Tara catalog, search/practice, discovery/launch, session, scheduling, selection/handoff, context/lineage, reflection, companions and ritual assembly/tone increments delivered; remaining room depth under review. |
| Shared platform                       | Contracts/data, communication/discovery, Sophia, Iris, Psyche/Lilith, trust/privacy and operational mechanics.                                          | Contracts/tenancy/persistence, events/jobs/discovery, Sophia depth and Iris depth delivered; Psyche/Lilith and wider trust/privacy/operations review remains active.                                               |
| Creation and workbenches              | Isis, Atelier, Studio, Living Scenes, Agentic Studio, shared kit and all seven workbench initiatives including Yemaya Study.                            | Living Scenes composition, techniques, compatibility, assist, governance and consumer/sharing audits delivered; remaining creation and workbench coverage requires expansion.                                      |
| Channels, commerce and administration | Telegram and other channels, Aje/crypto, fiat/store payments, support, tenant and operator work.                                                        | Pending full review and enhancement.                                                                                                                                                                               |
| Surface and design depth              | Web/PWA, native/watch, relevant real UI, design-system principles and particular interaction decisions.                                                 | Pending full review and enhancement.                                                                                                                                                                               |
| Operations and release                | Providers, runtime composition, observability, tests, deploy/rollback, recovery and acceptance evidence.                                                | Pending full review and enhancement.                                                                                                                                                                               |
| Organization and final quality        | Prerequisites, routes, source/slide assignments, every-guide sequence and every-slide optical review.                                                   | Pending integrated review.                                                                                                                                                                                         |
| Narration, PDFs and publication       | Current narration, final PDFs, relevant automated gates and retained evidence; normal hooks and task tip on origin/main.                                | Baseline current; every change requires affected delivery evidence.                                                                                                                                                |

No workstream is closed by a guide count, layout statistic or citation match.
Follow the applicable authoring and verification requirements in `TODOS.md`
sections 0, 7, 10 and 11. Human reader/listening evidence must be labeled
accurately; unperformed sessions cannot be counted as acceptance. Continue
independent work while any such feedback is pending.

## First package: member foundation

Audience: product/design/engineering readers who understand the V1 release line.
Outcome: follow a member's choice into the supported surface and identify which
owner can confirm its result. Preserve the eleven stable web/PWA slide IDs and
existing deep links; add separate explanations when necessary.

Review sources: V1 brand, product-promise, product-surfaces, account/profile and
design-system feature/architecture pages; actual navigation release scope,
registry consumers, PWA/update/offline helpers, account routes and their tests.
Record contradictions explicitly. A source's broad platform inventory does not
make a room or native surface part of the current release.

Deliver canonical manuscript/diagrams, source review and slide briefs, affected
HTML/PDF/narration/catalog outputs, optical evidence and source assignments
where complete unit review supports them. Review the actual candidate before
generating new speech. Use existing focused unit/source/Playwright/PDF checks
and add useful semantic regressions. Reserve the full shared-renderer/bank check
for a settled shared change; avoid rerunning unchanged broad gates during
authoring.

Recovery: all work is isolated in `codex/v1-presentation-completion-20260908`.
Stop only owned processes, retain logs and valid partial narration output, and
keep unfinished guides in draft state until their delivered artifacts pass.
Publication of a verified increment does not close this full V1 objective.

## Foundation delivery checkpoint

The
[twelve-slide foundation receipt](../verification/v1-foundation-2026-09-08/README.md)
records the first completed authoring increment, all six affected guide
contexts, PDFs and the single new recording. The incoming Eve configuration
explanation from `431faf76eeb7e64deeb927b367f65113ac3375aa` is preserved. The
integrated library has 580 unique slides; the full V1 objective remains active.
Next deepen account/profile and offline mechanisms, then the four room feature
sets.

## Account delivery package

The [dedicated account guide](v1-account-controls.md) follows the member’s push
preference change through the hub, projection, persistence and specialist
readback. Seventeen slides add payload specimens, a local refresh gate, avatar
cleanup, memory action boundaries, voice/safety, notifications, Telegram, data
rights, connection and billing explanations. All seven account-feature units
have explicit reviewed teaching assignments; supporting source families remain
scoped to the cited claims. The
[account receipt](../verification/v1-account-2026-09-08/README.md) records
narration, the final browser/PDF review and publication inputs. This package
does not close offline, room or full V1 coverage.

## Offline continuity delivery package

The [offline guide](v1-offline-continuity.md) adds twenty slides about installed
reading, worker activation versus page reload, route/scroll restoration, cache
classification and freshness, document fallback, recent-route warming, explicit
web queueing and HTTP replay, reusable library storage/retry/order/replay, crash
boundaries and scoped session clearing. Four complete PWA/offline documentation
units have reviewed teaching assignments, with old manifest/common-queue/flush
claims reconciled against pinned code. The
[receipt](../verification/v1-offline-2026-09-08/README.md) retains final
browser, PDF, narration and source-fixture evidence. The integrated library has
620 unique slides, preserving the three incoming Eve operating-state slides.
This closes the technical delivery increment, not human acceptance or full V1.
Tara’s existing ten-slide guide still needs practice, session, scheduling,
player, companion and governed mentor depth; its content workbench belongs in
the broader workbench expansion. The complete Tara feature and workbench
overview have been read for discovery, with underlying contracts/callers still
under review and no full Tara source-unit acceptance claimed.

## Tara session delivery package

The [session guide](v1-tara-sessions.md) adds 21 slides about lifecycle history,
separate session/domain/player completion contracts, continuation readiness,
skip, recovery, completion projections and routing continuity. Four complete
feature units have reviewed teaching assignments. Two overview explanations were
corrected against actual schema behavior. The
[receipt](../verification/v1-tara-sessions-2026-09-08/README.md) retains final
browser/PDF, narration and source-characterization evidence. The library has 641
unique slides. This closes the session delivery increment, not all Tara or V1
work. [Scheduling](v1-tara-scheduling-discovery.md) and
[practice surfaces](v1-tara-practice-discovery.md) have substantive source
discovery; player/checkpoint, companion, mentor and workbench review remain
required, alongside all other open workstreams above.

## Tara scheduling delivery package

The [scheduling guide](v1-tara-scheduling.md) adds fifteen slides after the
session guide: declared policy versus consumed decisions, continuity timing,
unique and repeated/skipped local clocks, quiet-hour windows and active
weekdays, primary and follow-up reminders, runtime versus injected time, cloned
in-memory storage, due-query eligibility and reschedule identity. Its
[receipt](../verification/v1-tara-scheduling-2026-09-08/README.md) retains the
final 111-check layout matrix, complete focused center/PDF checks, fifteen
recordings and sixteen source-characterization cases. The 659-slide library
preserves all 644 earlier slides and recordings. No whole feature unit is newly
assigned: the combined ritual/steps/scheduling unit still needs its remaining
practice/assembly explanation. The expanded
[practice discovery](v1-tara-practice-discovery.md) distinguishes the web
player, local reflection states and the separate shared audio/checkpoint model.
Full Tara, other rooms, all wider V1 work and human acceptance remain open.

## Tara web practice delivery package

The [web practice/player guide](v1-tara-web-practice.md) adds 21 slides
immediately after the Tara overview, before session and scheduling depth. It
follows the real course rows and controls, local transitions, the 80% finish
boundary, text/ambient/voice separation, guarded start requests, completion
failure cases, sequential source/canonical writes, reflection behavior and later
room/streak readback. Three focused product captures are backed by thirteen
controlled source-component cases; ten immutable-source VM cases characterize
room and repository behavior. The
[receipt](../verification/v1-tara-web-practice-2026-09-08/README.md) retains 153
visual checks, complete focused center/PDF checks and 21 recordings. All 659
previous slides and recordings remain exact. No whole feature unit is newly
assigned: catalog/recommendation, practice assembly and native callers still
require their owning explanation and review. The full V1 program, other rooms,
companions, mentors, workbenches and human acceptance remain active.

## Tara catalog delivery package

The [22-slide catalog guide](v1-tara-practice-catalog.md) follows linked
practice, content and teacher construction before taxonomy, duration, rights,
accessibility, media, cadence and the separate contract boundary. Twenty
immutable-source cases supply worked outputs; twelve cited or loaded source
files match the pin and implementation baseline. The
[receipt](../verification/v1-tara-catalog-2026-09-08/README.md) retains 160
final visual checks, full focused center/PDF checks and 22 recordings. All 680
earlier slides and recordings remain exact. No combined feature unit is closed
by these citations. Actual recommendation/search/saved/launch consumers,
complete taxonomy selection policy, native callers, companions, mentors and
workbenches remain required, alongside the full V1 program.

## Tara catalog and Eve service-level integration

The incoming Eve operations change adds four service-level slides while the Tara
catalog delivery adds twenty-two. Both remain complete in the merged library: 66
guides, 706 unique slides, 729 guide-slide instances and 403 cited paths. Every
guide HTML file and all 706 recordings remain exact to their owning delivery.
The
[integration receipt](../verification/v1-tara-catalog-integration-2026-09-08/README.md)
records current focused center/PDF checks and preservation across both parents.
Search/launch and taxonomy-selection discoveries prepare the next required Tara
subjects; all remaining V1 workstreams stay active.

## Tara web search delivery package

The [sixteen-slide web search guide](v1-tara-web-search.md) follows the actual
inline result pool through debounce, one-pass synonyms, weighted substring
scores, fixed type sections, original highlights, spelling and empty results.
History events, exact-string identity, failed storage and supported route
requests receive independent explanations. Nineteen complete source-component
cases and two readable source crops support native inventory/history/branch
diagrams and worked comparisons. The
[receipt](../verification/v1-tara-search-2026-09-08/README.md) retains 117 final
visual checks, focused center/PDF checks and sixteen recordings. All 706 earlier
slides and recordings remain exact. This does not establish a live catalog,
destination rendering or new whole-unit acceptance. Canonical
recommendation/search/saved consumers, full taxonomy selection, native callers,
companions, mentors, workbenches and all other open V1 work remain required.

## Tara service discovery delivery package

The [twenty-four-slide guide](v1-tara-service-discovery.md) explains direct and
canonical reads, BFF scope admission and role declarations, configured pools,
filtering, result identity, freshness and bridge ordering. It follows favorite
intent and authenticated route calls into compact sync records, complete versus
partial reconciliation, raw-page completeness, duplicate handling and failure,
then action mapping, path/query precedence and parsed intent. Twenty-six source
cases exercise complete modules with explicit service, middleware and clock
seams; twenty-two cited or loaded paths have pinned byte continuity.

The [receipt](../verification/v1-tara-discovery-2026-09-08/README.md) binds
native diagrams, 174 final visual checks, complete focused center/PDF checks and
24 new recordings. All 722 prior slides and recordings remain exact. This
increment does not close a whole Tara source unit or the full V1 program.
Application saved-item consumers, storage/retry, actual destination content,
native/audio, full selection and crisis policy, companions, mentors, workbenches
and every other open workstream remain required.

## Tara selection and handoff delivery package

The [linked selection and handoff guides](v1-tara-selection.md) add 44 slides.
The first covers all twelve moods, thirty-seven slate entries, fifteen themes
and fourteen modalities, then effective distress, independent compatibility,
cadence, constraints, fallback metadata and validator boundaries. The second
traces classifier normalization, text/context admission, declared detector
order, explicit type precedence, divergent trigger/reason fields, all thirteen
rule summaries and resource planning. Native relationships, a priority ladder,
branch diagrams and field specimens accompany full references and worked cases.

The [receipt](../verification/v1-tara-selection-2026-09-08/README.md) retains 64
unmodified owning tests, 21 additional source cases, 316 final visual checks,
focused center/PDF checks and 44 new recordings. All 746 prior slides, scripts
and recordings remain exact. These source mechanisms do not close a combined
Tara unit or full Lilith acceptance. Application selection/start enforcement,
resource delivery, recovery, native/audio, saved-item consumers, companions,
mentors, workbenches and all other open V1 workstreams remain required.

## Tara selection and Eve release integration

The combined library retains both linked Tara guides and the incoming Eve
release-evidence page: 70 guides, 791 unique slides, 814 guide-slide instances
and 416 cited paths. Every guide HTML file, both manuscripts and all recordings
remain exact to their owning delivery. The
[integration receipt](../verification/v1-tara-selection-integration-2026-09-08/README.md)
records focused center/PDF checks for all three guides and preservation across
both parents. The incoming computer-use and Docs Center changes are retained.
The remaining V1 program, actual Tara application consumers and native/audio,
companions, mentors, workbenches and integrated organization remain active.

## Tara selection and complete Eve operations integration

Main advanced during publication with one final Eve operations reading-map page.
The
[final integration receipt](../verification/v1-tara-selection-final-integration-2026-09-08/README.md)
retains both parents, all 792 recordings, all 70 guide HTML files, unchanged
Tara exports and a current twenty-page Eve review. The library now has 815
guide-slide instances and 416 cited paths. The incoming source-scoped Eve
operations completion does not close the remaining full V1 program.

## Tara shared audio and native playback delivery

The [audio pair](v1-tara-native-audio.md) adds 23 shared-state slides and 19
native-playback slides. It teaches the full command matrix, interruption,
engine-update precedence, wall-clock listening, completion, marker intervals,
event and snapshot ownership, then actual native metadata, attachment, launch,
status, controls, readiness mismatches and cleanup. The
[receipt](../verification/v1-tara-audio-2026-09-08/README.md) binds 120 owning
tests, 19 audio source cases, nine exact native-caller cases and final visual,
accessibility, PDF and narration evidence. All 792 prior recordings remain
exact.

This source-scoped delivery does not close the whole Tara source unit or native
acceptance. Offline files/checksums, checkpoint and reflection persistence,
quick reset, sleep, companions, mentors, workbenches, application start/handoff
and saved consumers remain required, alongside all other open V1 workstreams.

## Tara selection source-unit assignment

The existing selection guide now binds seventeen pages to the five complete
taxonomy units in `V1/features/domain-tara.md`: the canonical-catalog boundary,
mood, theme, modality and breathwork-cadence sections. Each assignment retains
the exact unit heading, source-unit hash, guide/slide identity and enriched
slide fingerprint. The source-scoped inventory advances to ten reviewed, two
structural and eighteen unassigned units with no stale-assignment errors; global
reviewed coverage advances from 83 to 88.

The
[assignment receipt](../verification/v1-tara-selection-assignment-2026-09-08/README.md)
retains the exact regression, 316 current browser checks, a focused center run
against the integrated baseline and the complete 25-page PDF review. The library
remains at 70 guides, 792 unique slides, 815 guide-slide instances, 416 cited
paths and 792 unchanged recordings. The assignments explain the pinned catalogs
and their procedural boundaries; they do not establish a live member assessment,
application enforcement, clinical acceptance, resource reachability or human
comprehension. The remaining Tara and full V1 work stays active.

## Tara audio and selection integration

The
[current integration receipt](../verification/v1-tara-audio-integration-2026-09-09/README.md)
combines the 42 audio pages with the incoming five selection-taxonomy
assignments. All 72 guide HTML files, 834 narration tracks and both
source-review histories remain attached to their owning revisions. Current
center review passes fourteen checks; all 42 re-exported audio PDF pages match
the completed optical review pixel for pixel. The manuscript preservation
baseline now includes incoming main's exact source-heading citations. No source,
human or native acceptance claim is expanded by this merge, and all remaining V1
work stays active.

## Tara offline audio and retention

The [twenty-slide guide](../decks/tara-offline-audio.html) follows native
playback through the cache and download manager. It explains the three records,
checksum representation, presence-media fallback, quality replacement, recency
ordering, operation-triggered expiry, budgets, accounting and file-cleanup
boundaries. The [receipt](../verification/v1-tara-offline-2026-09-09/README.md)
binds ten unmodified owning tests, seventeen source cases, current
browser/center/export review and twenty new recordings. All 834 earlier slides,
scripts and audio entries remain exact. These source-scoped explanations do not
close the entire Tara handbook unit, native storage/playback acceptance or the
rest of V1. Reflection, checkpoint consumers, reset/sleep, companions, mentors
and workbenches remain required alongside the other open workstreams.

## Tara duration, context and lineage delivery package

The [field guide](v1-tara-context-lineage.md) adds 24 slides after the linked
selection and handoff guides. It preserves every duration bucket with literal
range, cadence, default and spacing values; all ten context records with local
bands, signals, quiet policy, buckets and theme bias; and all eight lineage
records with families, named teachers, references, disclosure and review
metadata. Executed examples retain the 12-minute availability edge, wrapping
Night band, catalog-order signal result, quiet-hours override, cross-lineage
persona gate, declared-frame mismatch and unknown-lineage refusal.

The [receipt](../verification/v1-tara-context-lineage-2026-09-09/README.md)
retains 14 pinned source cases, 172 browser checks, all 24 optically reviewed
PDF pages, 24 new recordings and exact reuse of the 854 baseline scripts and MP3
bytes. Three assignments bind nineteen cited pages to the duration, context and
lineage handbook units. The source-scoped report now has thirteen reviewed, two
structural and fifteen unassigned units; global reviewed coverage is 91. The
library has 74 guides, 878 unique slides, 901 guide-slide instances and 427
cited paths.

The feature helpers return candidate records and policy results. Their bounded
direct-symbol search did not locate an application caller, so the combined
worked case leaves content choice, disclosure, launch and review with an
unresolved owner. Native/audio behavior, recommendation assembly, companions,
mentors, workbenches, other member rooms, cultural and rights acceptance, human
comprehension and every other open V1 workstream remain active.

## Tara reflection and continuation

The [twenty-four-slide guide](../decks/tara-reflection.html) closes the selected
native reflection teaching gap: entry and context precedence, all five prompt
families, capture transitions, completion conversion, timestamp ownership,
recent-rail policy, route identity and local storage boundaries. The
[receipt](../verification/v1-tara-reflection-2026-09-09/README.md) records
eleven unmodified native owning tests, forty-three library owning tests,
nineteen source/caller cases and current browser, center, PDF and narration
evidence. All 854 earlier canonical slides, scripts, audio entries and bytes are
preserved. The native screen is audited through named declarations, not asserted
to be fully reviewed. Device lifecycle, durable reflection recovery, checkpoint
consumers, reset/sleep, companions, mentors, workbenches and other V1 work
remain open. Incoming Eve watcher-semantics work at 6d003eeda1 is integrated
without changing or claiming presentation coverage for those new source units.

## Combined reflection and context/lineage delivery

The
[integration receipt](../verification/v1-tara-reflection-integration-2026-09-09/README.md)
preserves the reflection leaf at `3999253c302734970cc1ae9e632e9b70d33ec658` and
incoming context/lineage delivery at `dec34aacff0f40e368e5caf412d5df43df4a6a41`.
Both twenty-four-slide guides keep their complete manuscripts and recordings.
The combined library has 75 guides, 902 unique slides, 925 guide-slide instances
and 429 cited paths; all narration is current. Fresh combined-center checks and
126 targeted presentation regressions pass. The 24 reflection PDF pages retain
the exact pixels of their owning review. The 24 context/lineage pages receive a
new optical review because the incoming Linux export and current macOS export
render different typography. All 75 guide HTML files retain their parent bytes.
Both original receipt trees remain unchanged historical evidence. The incoming
assignments bring Tara to 13 reviewed, 2 structural and 15 unassigned units;
global reviewed coverage is 91. Native-device and human acceptance, remaining
Tara subjects, the other V1 workstreams and integrated organization remain open.

## Tara companion selection and handoffs

The [twenty-eight-slide guide](../decks/tara-companions.html), after reflection,
explains all six native offers, moment precedence, Nisaba/Arete/Nyx mappings,
Metis topic scoring and excerpt handling, Veritas/Sophia authored evidence
metadata, Assistant session context, access guards and navigation boundaries.
The [receipt](../verification/v1-tara-companions-2026-09-09/README.md) records
29 native and 30 library owning tests, nineteen source/caller cases, complete
browser/PDF/narration review and preservation of 902 prior slides and
recordings. The complete Cross-domain companions handbook unit is now reviewed
with eight bound teaching slides. Tara is at fourteen reviewed, two structural
and fourteen unassigned units; global reviewed coverage increases by one to
ninety-two.

This is a presentation delivery, not a product repair. The native Metis memo
omits reflectionText; the basic three companions ignore it; explanatory labels
and source summaries are authored; the Assistant handler prepares context and
emits its saved notice without a direct Iris write. The sheet request path was
traced in source, not executed against the BFF. Native mounting, destination
release/access/entity results, durable memory and restart recovery need their
own evidence. Tara checkpoint consumers, reset/sleep, saved-item consumers,
mentor presence, workbenches, destination-wide feature audits, the remaining V1
areas, integrated organization, human listening/comprehension and native-device
acceptance remain open. Source citations do not close unreviewed units.

Incoming Eve skill-system reconciliation at `315d53f002` is integrated. Both
presentation trees and every audited source file are unchanged by that commit;
the
[continuity record](../verification/v1-tara-companions-2026-09-09/integration.json)
records the comparison. This increment does not claim coverage of the new Eve
skill-reconciliation behavior.

## Tara teacher profiles, lineage and actions

The twenty-two-slide teacher guide — since merged into
[chapter five of the Tara track](../decks/tara-selection.html) — follows the
duration/context/lineage guide in the V1.0 route. It deepens identity, canonical
lineage, roles, locale coverage, rights, accessibility and practice references,
then follows the actual directory through search, limited specialty controls,
status, lineage uncertainty, credentials, content joins, generated reviews,
weekday availability, enabled actions and unknown-ID recovery. Two actual React
component crops and source-shaped relationships, records and gates support the
teaching. The [receipt](../verification/v1-tara-teachers-2026-09-09/README.md)
records eleven contract/helper cases, eight owning component tests, twenty
browser component checks, final guide/center/PDF review and narration. All 930
earlier canonical slides and recordings remain exact.

The complete Teachers handbook unit now has seven current teaching assignments.
Its de-duplication wording is reconciled with code that rejects duplicates. Tara
has fifteen reviewed, two structural and thirteen unassigned units; global
reviewed coverage increases to ninety-three. Source citations alone do not close
other units. The directory uses simulation data rather than the canonical
schema, its optional lineage display is a separate model, and null consent can
pass the contract. Generated reviews and weekday checks are not customer
evidence or bookings. No application behavior was changed.

Mentor Presence and Living Scenes remain next source families; their registry,
release, choreography and web/native caller relationships need separate
teaching. Tara checkpoint consumers, reset/sleep, saved-item consumers,
workbenches, all other V1 room/platform/creation/channel/operations audits,
integrated organization, human listening/comprehension and native-device
acceptance remain open. The full V1 goal is still active.

## Tara ritual assembly and tone review delivery package

The [thirty-slide field guide](v1-tara-ritual-assembly.md) follows the six
declared ritual moments from an input record to assembled steps, then tests
anchor precedence, derived and refused ingredients, journaling overrides and
duration budgets. It keeps the local domain template, feature publishing record
and shared contract visibly separate before tracing authored script
fingerprinting, literal tone-context precedence, eight deterministic quality
axes and current-review invalidation.

The [receipt](../verification/v1-tara-ritual-assembly-2026-09-09/README.md)
retains 20 executed source cases over 17 pinned files, 214 browser checks, all
30 optically reviewed PDF pages, 30 new recordings and exact reuse of the 878
baseline scripts, manifest entries and MP3 bytes. Two assignments bind 17 cited
pages to the ritual/steps/scheduling and contemplative-tone handbook units. The
Tara source now has fifteen reviewed, two structural and thirteen unassigned
units; global reviewed coverage is 93. The library has 75 guides, 908 unique
slides, 931 guide-slide instances and 438 cited paths.

The implemented assembly can return a protected core above the requested
duration and can remove an optional-kind step even when that step is marked
required. The feature builder accepts a nonblank modality outside the canonical
contract taxonomy, and its tone-review record does not carry the separate
tradition-consent decision. A bounded direct-symbol search found no converter or
application consumer joining the three template records. Application launch,
editorial and tradition authority, rights and cultural acceptance, member
comprehension, normal-speed listening, the remaining Tara units and every other
open V1 workstream remain active.

## Tara ritual, reflection and companion integration

The
[current integration receipt](../verification/v1-tara-ritual-assembly-integration-2026-09-09/README.md)
preserves the 930-slide incoming main at `2577f292de` and the thirty-slide
ritual-assembly delivery at `6b2b667130`. The merged library has 77 guides, 960
unique slides, 983 guide-slide instances and 460 cited paths, all with complete
narration. Tara now has sixteen reviewed, two structural and twelve unassigned
handbook units; global reviewed coverage is 94.

The reflection, companion and ritual packages retain their own scoped source,
browser, PDF and optical evidence. The integrated checkpoint proves current
generated outputs, all 960 audio files and the unchanged parent histories; it
does not expand their application, native-device, tradition/rights or human
acceptance claims. The remaining Tara and full V1 work stays active.

## Teacher and ritual integration checkpoint

The teacher guide — since merged into
[chapter five of the Tara track](../decks/tara-selection.html) — is integrated
with the concurrent
[ritual assembly and tone guide](../decks/tara-ritual-assembly.html). The
[combined receipt](../verification/v1-tara-teachers-integration-2026-09-09/README.md)
preserves the original receipts and all parent narration. Together they add 52
slides to the shared 930-slide baseline; the library has 78 guides and 982
unique slides. Three additional complete Tara handbook units are reviewed:
Teachers; Rituals, steps, triggers, and scheduling; and Contemplative tone
review. Tara totals are 17 reviewed, 2 structural and 11 unassigned, with global
reviewed coverage at 95. Citations do not close other units.

The combined review caught one incoming five-row contract table overflowing at
1440×900. Five shortened row labels retain the rejected examples, exact issue
paths, notes and narration; its single assignment fingerprint is refreshed. All
other canonical slides and every recording from both parents are preserved. The
teacher PDF’s 22 pages match its reviewed leaf pixels; the 30 current ritual
pages receive a fresh optical review. The final guide/center/PDF checks and 128
presentation regressions pass. No human, native-device or connected-service
acceptance is claimed. Mentor Presence, Living Scenes, checkpoint and saved-item
consumers, reset/sleep, workbenches, all remaining V1 domains/platform/creation/
channels/operations and integrated organization remain active.

## Tara Mentor Presence: score, choreography and actual delivery

The [43-slide guide](../decks/tara-mentor-presence.html) introduces the current
web result and preferences, then the cast and score before their constraints.
Settings, arrivals, posture, cadence, state transitions and crisis decisions
lead into motion and release review. The second half follows actual rendering,
provenance, ownership, durability, fallback, disclosure, events and aggregate
analytics. Three actual component captures use controlled network replies and
real development-render media. The guide sits under V1.x in Creation systems,
with teacher profiles and ritual assembly as prerequisites.

The [receipt](../verification/v1-tara-mentor-presence-2026-09-09/README.md)
records 27 source cases, ten unmodified owning web tests and twelve actual
component characterizations. Twelve complete handbook units have current
teaching bindings: one Domain Tara section and eleven Mentor Presence sections.
Tara has eighteen reviewed, two structural and ten unassigned units. The
standalone dependency survey, broad governance and complete accessibility units
remain unassigned; structural headings there also remain open. All 982 prior
canonical slides, scripts, manifest records and recording bytes remain exact.

The executed renderer produces a tonal APNG rather than an embodied mentor. Its
release report is separate from the condition that currently admits rendering.
The guide also explains the generic caller context, separate APNG clock,
nullable counted cadence, incomplete performance identity, disclosure
measurement gap and native caller uncertainty. Browser cases reproduce late
attachment after opt-out and image decode failure remaining active. These are
characterized product gaps; this delivery changes the presentation, not the
application.

The Living Scenes consumer audit is delivered below. The remaining Tara consumer
paths, other V1 domains, platform, creation workbenches, channels, operations,
complete sequence review, native-device acceptance and human
listening/comprehension remain open. The full V1 goal stays active.

## Living Scenes correction checkpoint

The [runtime correction package](v1-living-scenes-corrections.md) replaces five
inaccurate or compressed explanations and adds three detail slides inside the
existing guide. The
[receipt](../verification/v1-living-scenes-corrections-2026-09-09/README.md)
records the 16-page delivered guide, eight recordings, source cases and visual
review. All 1,020 unchanged slides and recordings are preserved. No complete
Living Scenes unit is newly assigned. Next audit actual composition, player,
rendering, reproduction and distribution consumers, alongside the remaining Tara
and wider V1 work. Full V1 and human acceptance remain open.

## Living Scenes overview correction checkpoint

The [remaining overview fixes](v1-living-scenes-authoring-fixes.md) update four
stable slides across their two canonical manuscripts. Responsibility comparisons
replace numbered peers, policy and promotion show explicit rejection branches,
and the three authoring surfaces have distinct records and execution limits. The
[receipt](../verification/v1-living-scenes-authoring-2026-09-09/README.md) binds
the sixteen-slide guide, four changed recordings, source cases and current Eve
defect checks. No complete handbook unit is newly assigned. Composition, full
authoring, templates, rendering, sharing and the wider V1 audit remain active;
these corrected overview pages do not replace those deeper source units.

## Living Scenes composition and assist checkpoint

The [composition depth package](v1-living-scenes-composition.md) adds nineteen
slides for sources, tiers, crisis admission, arranging, outgoing technique
ownership, text preview, accept/cap behavior, promotion, fallback maps,
direction and web/native differences. Twelve connected assist slides cover
requests, candidate validation, policy filtering, budget signals, explicit
refusals, gold-set arithmetic and challenger comparison. Five actual component
crops and two adjacency diagrams make the member workflow inspectable; six
decision diagrams expose both outcomes. All 1,028 earlier slides and recordings
remain unchanged. The library now has 81 guides, 1,059 slides and 502 cited
paths.

Three complete handbook units receive reviewed assignments with the
implementation qualifications taught directly. The helpers do not establish
pair-specific ranking, overlap rendering, automatic run cancellation, a
persisted full Score or model deployment. Native host testing remains separate
from native-device acceptance. The
[receipt](../verification/v1-living-scenes-composition-2026-09-09/README.md) and
[per-slide ledger](v1-living-scenes-composition-ledger.json) bind the current
artifacts and review scope.

The complete cinematographic catalog and primitive mechanisms, full Score
Editor, Tara venue publication, templates, render/sharing paths and all other
pending V1 workstreams remain open. This checkpoint does not close the full
objective or claim human comprehension or normal-speed listening acceptance.

## Living Scenes technique and compatibility checkpoint

The [technique depth package](v1-living-scenes-techniques.md) adds a twenty-five
slide catalog and continuity guide plus a nineteen-slide compatibility and
admission guide. The learning route now moves from the member composition
surface into all twelve catalog records and nine primitive contracts, through
general and technique-specific continuity evidence, then into pair policy,
seven-dimension scoring, the full twelve-by-five template matrix and bounded
override behavior before Compose Assist.

The visual system uses diagrams for dependencies, policy order and independent
measurement inputs; exact tables retain complete catalogs, matrices and numeric
bounds. Two worked cases show the five-field scorecard and a seven-value 0.70
compatibility result. Eleven complete handbook units have reviewed assignments,
and the package preserves all 1,059 earlier slides and recordings. The library
now has 83 guides, 1,103 unique slides, 1,126 guide-slide instances and 507
cited paths.

The slides qualify each contract at its real boundary. Primitive validation is
separate from DSP and shader execution. The catalog's outer freeze is shallow.
The main technique gate does not receive a tone band or enforce `requires` and
`forbids`. Four catalog techniques lack a technique-specific observation case.
Enforcement strings do not establish one-frame timing or application callers,
and the override helper does not establish durable persistence. The
[receipt](../verification/v1-living-scenes-techniques-2026-09-09/README.md) and
[per-slide ledger](v1-living-scenes-techniques-ledger.json) bind the final
artifacts and review scope.

Full Score Editor, Tara venue publication, downstream renderer and measurement
acceptance, distribution/sharing and the remaining V1 workstreams remain open.
This checkpoint does not claim human comprehension or normal-speed listening
acceptance.

## Living Scenes consumers, sharing and residual audit

The [47-slide guide](../decks/living-scenes-consumers.html) follows the customer
card and fixture-backed compose workspace into keep, sharing, public resolution,
render admission, durable media, provenance, accessibility and distribution. It
then audits policy ownership, release engines, the score editor, cue controls,
tier caps and caller connections. Five actual UI captures cover premium, unwired
and denied card states plus desktop and mobile composition. The guide sits
beside the existing Living Scenes runtime and governance guides under the V1
Creation path.

The [receipt](../verification/v1-living-scenes-consumers-2026-09-09/README.md)
binds eleven pinned source cases across thirty owning files, focused runtime,
kernel, BFF and web checks, the existing card and compose journeys, final guide
and center review, and all 47 PDF pages. Forty complete source units are newly
reviewed through 61 evidence bindings. Global coverage is 150 reviewed, 10,018
structural and 103,176 unassigned units. The library has 82 guides, 1,106 unique
slides, 1,129 guide-slide instances and 534 cited source entries. All 1,059
scripts, manifest entries and recordings from the integrated remote revision
remain exact, along with all 81 integrated guide HTML files.

The current card honestly reports absent keep, share, gallery and branch
callbacks. Tenant policy is server-owned and standalone rendering now requires
identity, bounded inputs and a shared budget lease. Materialized APNG bytes
survive snapshots and can be redeemed. The compose workspace still edits local
fixtures, public rendering does not consume the saved score, crisis grants are
self-asserted by the caller, and the conductor, cue and release engines have no
application admission point. First-frame metadata targets a missing route, the
visible mark remains DOM-only, reduced motion still keeps the animated APNG, and
several accessibility assets remain internal URIs. Release SLOs and deployment
secrets have source-defined contracts without production evidence.

The remaining Tara checkpoint, reset/sleep and saved-item consumers, other V1
domains, shared platform, creation workbenches, channels, operations, complete
sequence review, native-device acceptance and human listening/comprehension
remain open. The full V1 goal stays active.

## Living Scenes integrated sequence checkpoint

The two independently reviewed additions coexist without changing their shared
1,059-slide baseline. Their 91 slide IDs and narration IDs are disjoint. The
combined learning order is runtime → composition → technique catalog →
compatibility → Compose Assist, with governance as the policy branch and the
consumer audit after both assist and governance. The library now has 84 guides,
1,150 unique slides, 1,173 guide-slide instances and 537 cited source paths.
Together the packages add 51 reviewed source units; global coverage is 161
reviewed, 10,018 structural and 103,165 unassigned units.

Integration also repaired two delivery defects in the incoming consumer package:
the manifest-only `living-consumer-secrets` narration now has its MP3, and the
customer-card product view uses a shorter headline that fits the compact desktop
reader while preserving the full capture and callouts.

The source characterization, browser captures, PDFs and narration evidence stay
in the two owning receipts. The
[integration receipt](../verification/v1-living-scenes-integration-2026-09-09/README.md)
establishes coexistence, ordering and preservation; it does not reinterpret
those scoped observations or claim full V1, native-device, production, human
comprehension or normal-speed listening acceptance.

## Tara consumer continuity checkpoint

The [34-slide guide](../decks/tara-consumer-continuity.html) follows a 2:20
interruption through the distinct canonical and generic web players, UI and BFF
completion states, local checkpoint envelopes, quick reset, sleep downshift,
accessibility preferences, breath-haptic delivery, saved outputs and owner-level
recovery. Three existing product captures accompany record anatomy, state
relationships, evidence ladders and decision matrices. The guide sits in the
member-experience sequence after ritual assembly and carries a complete PDF,
3,218 words of speaker notes and 35.2 minutes of narration.

The [receipt](../verification/v1-tara-consumer-continuity-2026-09-09/README.md)
binds fifteen executed source cases across 28 pinned files, focused domain, BFF,
web and mobile tests, the final standalone and embedded browser matrices, and
all 34 PDF pages. Ten complete handbook units receive current teaching
assignments through 53 evidence bindings. `V1/features/domain-tara.md` now has
all 28 meaningful units reviewed and two structural headings; global coverage is
171 reviewed, 10,018 structural and 103,155 unassigned units. The integrated
library has 85 guides, 1,184 unique slides, 1,207 guide-slide instances and 569
cited source entries.

The canonical player uses a local clock and ambient audio while exposing a
display-only speed value; the generic player owns real playback speed but keeps
favorites and reflection locally. The canonical completion projection replaces
partial UI state with a separate fixed completed record. Standalone Tara
persists checkpoint, Iris and Psyche shapes locally, and its cross-device
browser case begins after transport would have occurred. Breath timelines reach
authenticated compilation and both watch runtimes, but no phone product flow
calls the bridge method. Unified Library mixes real favorites with two fixed
saved-looking examples; mobile downloads real bytes while only carrying asserted
watermark metadata. These boundaries are taught as current behavior rather than
delivery claims.

No connected cross-device transport, durable canonical reflection write,
phone-to-watch invocation, deployed-service acceptance, live native-device run,
normal-speed human listening or reader comprehension was established. All other
V1 domains, shared platform, creation workbenches, channels, operations and the
complete sequence review remain active. The full V1 goal stays open.

## V1 contracts, tenancy, events, jobs and discovery checkpoint

The shared-platform route now adds a 29-slide
[contracts, tenancy and persistence guide](../decks/v1-contracts-tenancy.html)
after the existing architecture overview and a 34-slide
[events, jobs and discovery guide](../decks/v1-events-jobs-discovery.html) after
the existing communication overview. The route continues into Sophia, Iris and
Psyche/Lilith so readers establish contract, authority, durability and recovery
vocabulary before interpreting evidence, memory and live interaction.

The first guide follows a preference write through contract projections,
partial-result truth, domain ownership, role/scope admission, the bounded OAuth
contract, process-local idempotency, trusted tenant propagation, explicit
cross-region consent, DSR routing, persistence drift, tombstone/erasure and safe
partial recovery. The second distinguishes HTTP, WebSocket, gRPC, events and
assigned jobs; then teaches the actual Redis pub/sub plus TTL/ack/scheduler/
claim/dead-letter substrate, outbound delivery, BullMQ job state and SLA
evidence, and the live search/recommendation path from candidate pools through
ranking, feedback and offline evaluation.

Fifty-six meaningful documentation units are bound to exact slide fingerprints.
Nine pinned-source checks cover idempotency, tenant/residency, both event
envelope surfaces, event recovery primitives, queue state and worker defaults,
SLA metrics, live search weights/cache/ties and the recommendation scorer. That
review caught and corrected generic recommendation dimensions before release:
the live composite uses base relevance, diversity bonus, recency boost,
engagement alignment and preference alignment with fixed weights.

The [delivery receipt](../verification/v1-platform-depth-2026-09-09/README.md)
records the final browser, Center, PDF, source and narration evidence. The
integrated library has 87 guides, 1,247 unique slides, 1,270 guide-slide
instances and 605 cited source paths. Global coverage is now 227 reviewed,
10,018 structural and 103,099 unassigned units. This checkpoint deepens the
shared foundation; it does not close the remaining Sophia, Iris, Psyche/Lilith,
trust/privacy, operations, workbench, channel, commerce, domain, native-device
or human comprehension/listening work. The full V1 objective remains active.

## Sophia depth checkpoint

The [forty-slide Sophia guide](../decks/sophia.html) now follows one evidence
question through three explicit implementation postures. It begins with the live
grounded-answer and source-credibility routes, then inspects answer thresholds,
envelope fields, claim and state vocabularies, model fallback, scoring weights
and the 0.5 credibility floor. It continues through the versioned adapter, role
matrix, honest empty results, source-set readiness and trace export before
teaching the real-but-unwired BM25, fact-check and contradiction engines and the
specification-only ingestion, publication and invalidation architecture.

Thirty-two new slides use case studies, evidence comparisons, responsibility
maps, gated flows, recovery maps, source-record anatomy and exact decision
tables around the eight established slides. Four earlier reviewed units plus 49
new assignments cover every meaningful unit in the pinned Sophia architecture
and feature pages. Citation count remains an evidence-availability threshold,
source credibility remains separate from answer support, and implemented
libraries remain separate from a live customer-path claim.

The [delivery receipt](../verification/v1-sophia-depth-2026-09-09/README.md)
records 280 standalone and embedded visual assertions, 136 captures, four
rejected negative controls, a fresh 40-page PDF, 14 Center checks, 19
presentation tests and 14 executable pinned-answer cases. The 32 new narration
tracks add 42.6 minutes; the complete guide runs 56.5 minutes. The library now
has 87 guides, 1,279 unique slides, 1,302 guide-slide instances, 618 cited paths
and 28.61 hours of narration. Global coverage is 276 reviewed, 10,018 structural
and 103,050 unassigned units.

The current checkout cannot execute the historical BFF Sophia suites because its
shared dependency installation lacks `fastify` and the current tree no longer
contains the pinned `@oshun/evidence-sophia` workspace package; the retained log
records that import-time limitation. Canonical source snapshots, assignments and
executable pinned-answer tests preserve the V1 source boundary. No provider,
production, native-device, human comprehension or normal-speed listening
acceptance is claimed. Iris, Psyche/Lilith and the remaining V1 workstreams stay
active.

## Arete humane engagement checkpoint

The [43-slide guide](../decks/arete-humane-engagement.html) follows one partial
check-in through its validated record, humane policy, recovery stage, consumer
projection, streak fold, weekly review and Living Offering. It inventories the
goal, habit, routine, recovery, friction and reflection contracts before tracing
which states are durable, browser-local, component-local or fixture-backed. Four
actual route crops retain their visible unavailable or preview labels. The guide
sits after the Arete introduction and member account controls and carries a
43-page PDF, 2,654 words of speaker notes and 21.9 minutes of narration.

The [receipt](../verification/v1-arete-humane-engagement-2026-09-09/README.md)
binds 24 source cases across 30 pinned files, four current route
characterizations, focused contract, domain, BFF, web and mobile tests, 305
guide checks, 172 guide captures, the shared center run and all 43 PDF pages.
Twenty-two complete handbook units receive 50 exact teaching records. Together
with the earlier check-in assignment, `V1/features/domain-arete.md` has all 23
meaningful units reviewed and three structural headings. Global coverage is 298
reviewed, 10,018 structural and 103,028 unassigned units. The library now has 88
guides, 1,322 unique slides, 1,345 guide-slide instances and 642 cited source
entries.

The executed evidence exposes several boundaries. The consumer check-in row
compresses the canonical five-status event, and its BFF streak fold does not
consume per-habit policy. The deterministic recovery engine can calculate a
stage, but the consumer route does not persist the canonical missed-day,
recovery-record or recovery-plan spine. The weekly route mixes live review data
with fixture prompts, while the current recovery, review and Offering pages are
explicit previews. Mobile completion changes component state, telemetry and
haptics without posting the record. Shared concept threads remove Veritas and
Metis links because those rooms are deferred from V1; the corrected Arete test
now follows that existing release-scope contract.

No connected canonical recovery write, shared policy translation, mobile
transport, deployed-service acceptance, live native-device run, normal-speed
human listening or reader-comprehension acceptance was established. Nyx, Nisaba,
later domains, shared platform, creation workbenches, channels, operations and
complete sequence review remain active. The full V1 goal stays open.

## Iris depth checkpoint

The [forty-four-slide Iris guide](../decks/iris.html) now follows one fictional
member preference from a Psyche turn through the assistant shell, entry
identity, category, canonical and adapter scopes, governance policy, retention,
promotion, compaction and deterministic recall. The recall sequence keeps
tenant, scope, crisis, suppression, lifecycle, sensitive-context, ranking,
conflict, budget and audit behavior separate, including the corrected 30-day
exponential time constant and the difference between write precedence and recall
freshness.

The second half teaches entry consent, append and full-ledger validation,
inference authority, privacy suppression, delete/export/access requests, rights
deadlines and result fingerprints, the intended but partial member controls,
multi-actor masking, content-free continuation tokens, cross-device conflict,
inspection purpose and policy, replay, audit, notice and the shell bridge. The
sequence uses record anatomy, gated flows, ownership handoffs, evidence
comparisons, recovery maps and exact policy tables around the eleven established
slides.

Six established assignments plus 49 new assignments cover every meaningful unit
in the pinned Iris architecture and feature pages; one parent heading remains
structural. The
[delivery receipt](../verification/v1-iris-depth-2026-09-09/README.md) records
308 final browser/accessibility checks, 143 captures, four rejected negative
controls, 14 Center checks, 31 executable pinned-module cases and exact visual
text on the fresh 44-page PDF. The 33 new tracks add 38.37 minutes; the complete
Iris guide runs 61.51 minutes. With the concurrent Arete checkpoint integrated,
the library now has 88 guides, 1,355 unique slides, 1,378 guide-slide instances,
658 cited paths and 29.61 hours of narration. Global teaching coverage is 347
reviewed, 10,018 structural and 102,979 unassigned units.

The contracts and pure modules do not establish authenticated callers, durable
storage, replica and downstream erasure, audit ingestion, delivered member
notices, complete web/mobile controls or cross-device acceptance. No human
comprehension or normal-speed listening acceptance is claimed. Psyche/Lilith and
the remaining V1 workstreams stay active.
