# Portfolio and V2 orientation source review

This review covers the three original expansion guides:

- `portfolio-map` — 12 slides;
- `v2-product-promise` — 9 slides; and
- `v2-product-surfaces` — 12 guide instances, including the shared
  `v2-promise-shared-spine` slide.

That is 32 unique slides and 33 guide instances. Every explanation was read
against all of its cited sources at revision
`1ceae52c7896a59ac74a4429d410f3218f0b91c6`. The machine-readable
[review record](orientation-source-review.json) binds the assembled slide
fingerprints to 37 exact spans across all 27 cited paths. Each span carries the
pinned file hash, line range, range hash, literal anchors, and the reason it was
used.

This is a claim-level source review. It does not assign a complete Markdown
section as taught, does not change the teaching-coverage inventory, and does not
turn the portfolio map into detailed V2–V10 instruction. The catalog continues
to label it “Reviewed portfolio overview; detailed expansion in progress,” and
the center README continues to state that the overview and five V2 guides do not
establish comprehensive V2–V10 coverage.

## Findings

All 32 explanations are supported with recorded qualifications. No authored
slide, visual, PDF, or narration change is required.

| Guide area              | Source finding                                                                                                                                                                                                                                   | Preserved explanation boundary                                                                                                                |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------- |
| Portfolio identity      | The Docs Center registry names ten product spaces. The older platform overview still describes V1–V9, while the V10 feature and architecture sources establish the Rail as the tenth product that reuses V1 identity and platform substrates.    | The opening slide combines those sources and treats the portfolio as ten products sharing foundations; it does not claim equal maturity.      |
| V1                      | `V1/features.md` and `V1/BRAND.md` distinguish the whole V1 line from V1.0: V1.0 admits Tara, Arete, Nyx, and Nisaba on web/PWA; native/fiat belongs to V1.1 and Veritas/Metis to V1.2. Eve’s handbook keeps Eve on builder/operator planes.     | The V1 overview names the four admitted rooms and separates member-facing Lilith from builder-facing Eve.                                     |
| V3                      | The overview defines one embodied world with Tara Studio, Saraswati Stage, and Lilith Commons, then separates real machinery, content targets, client tiers, consent, and provider-dependent paths.                                              | The slide remains an experience and boundary map; inventory targets do not become proof of released rooms or qualified content.               |
| V4                      | The feature index presents five player-facing ruleset cells. The deeper product source also discusses a broader module/cell implementation inventory.                                                                                            | The slide says “several genre cells” and names representative experiences, avoiding a false single count across the player and engine layers. |
| V5                      | The product source and feature index consistently define six open-world narrative ruleset cells, tied together by shared player truth and the Mind Palace.                                                                                       | The slide preserves six cells and distinguishes source modules, plugin activation, authored content, and an assembled playable experience.    |
| V6                      | The sources support stewardship, persistent Ori identity, Chronicle continuity, refusal/counter-offer behavior, tiered clients, Studio, and Aye portability. They also mark live cognition, online, and GPU planes as provider/operations gated. | The slide presents the intended experience while retaining the provider and operational boundary.                                             |
| V7                      | The sources support server-authoritative realms, executable UGC, capability limits, reproducible dependency resolution, and governance, while naming integration and production gaps.                                                            | The slide does not infer a complete realm lifecycle from Rust services, a sandbox, or a lock-file composer alone.                             |
| V8                      | The sources separate the symbolic case skeleton and unique-solution proof from prose/media realization and the reused V5 player surface.                                                                                                         | The slide keeps fair solvability and complete player realization as separate acceptance obligations.                                          |
| V9                      | The sources support Atlas-scoped wonders, solve-first ground truth, versioned Lesson artifacts, computed domains, Threads, and mastery feedback.                                                                                                 | The slide does not treat a passing artifact structure as evidence of teaching effectiveness.                                                  |
| V10                     | Both sources label the Rail concept-stage. The feature source explicitly states that a global budget keeps ten subscribed channels manageable; the architecture assigns attention policy to the Rail and content ownership to channels.          | The slide retains concept status and requires integration/runtime evidence before support is claimed.                                         |
| Cross-product authority | The platform request trace and V3/V7/V8/V9/V10 sources consistently compose shared mechanisms without transferring the originating product’s outcome authority.                                                                                  | The closing portfolio slide remains a reading method, not a claim that one surface proves another product’s result.                           |

## V2 product promise

The V2 feature, architecture, and documentation sources agree on the intended
seven-tradition fighter, the shared product spine, tiered network authority,
complete-content obligation, audience-specific behavior, provider-gated
cross-platform support, failure behavior, and the evidence ladder. The slides
also retain limitations that the sources themselves require:

- the 60 Hz, eight-frame, 100 ms rollback envelope is a target, not a measured
  release result;
- a mode registry, planned GameFeature entry, editor surface, or docs checker
  cannot prove complete authored content;
- cross-play and cross-progression need named platform pairs, credentials,
  service configuration, and durable outcomes;
- accessibility and locale controls need verification in the supported device,
  modality, ruleset, and gameplay path; and
- contracts, checkers, modules, and workflow files make enforcement concrete but
  do not substitute for a current passing release result.

## V2 surfaces

The named CommonUI, `V2Modes`, web, developer-portal, service, and broadcast
artifacts support the surface explanations. The review preserves the
distinctions among the shell-native Next route, the standalone Vite/static web
family, public projections, operational records, and deterministic match
authority.

The companion explanation required the closest comparison. The feature overview
calls the planned mobile client “read-only,” but the pinned OpenAPI contract
contains scoped POST operations for coach metadata, notification subscriptions,
tournament operations, leaderboard ingestion, callbacks, and micro-game reward
validation. The existing slides already interpret “read-only” as a boundary on
deterministic gameplay authority rather than an absence of HTTP writes. They
also record the contract details that qualify the overview:

- the three read routes allow an anonymous security alternative as well as OAuth
  scopes;
- public visibility, owner-authorized private reads, and replay reads remain
  distinct access decisions;
- `gameplayAuthority` is fixed to `false` for the coach-stream response;
- the notification contract includes Web Push in addition to APNS and FCM; and
- rate-limit values in prose or response headers are policy descriptions, not
  proof that all limits are enforced by a running service.

No iOS or Android client exists at the pin. The live API contract can therefore
support design and integration work without being presented as a shipped mobile
application.

## Reproducible check

Run the source-review checker from the repository root:

```sh
python3 tools/presentations/check-orientation-source-review.py
```

The checker fails if the source revision, a pinned file or span, a literal
anchor, an assembled slide fingerprint, guide membership/order, cited-path
coverage, catalog status, or README scope boundary changes. Its regression suite
also injects those failures. A later intentional source or slide change requires
a new editorial review and refreshed evidence rather than a hash-only update.
