Status: Accepted
Date: 2026-09-11
Authors: Isis Platform, Oshun BFF, Studio
Reviewers: @oshun/core-team, @oshun/trust-safety, @oshun/release-eng
Formalises the decision recorded against 182.C.01.11 and required by
182.C.24.05. The tracker's prohibition 4 makes this a precondition: no
182.C.23 or 182.C.24 item may be checked before this decision exists.
Context and Problem Statement#
Phase 182's creator flow needs a surface. The item names three candidate desktop hosts, and all three exist:
| App | Runtime | Source files | What it is |
|---|---|---|---|
apps/yemaya/studio-desktop |
Electron (forge.config.ts, makers/) |
63 | "creative suite for film and game production" |
apps/lilith/desktop |
Tauri 2 (src-tauri/) |
14 | "consciousness experience platform" |
apps/iris/desktop |
Tauri (Playwright e2e present) | 19 | "AI Assistant" |
No Oshun studio desktop app exists. The release train
(libs/oshun/navigation/src/release-scope.ts) ships V1.0 web + PWA, V1.1
native mobile, V1.2 restores Veritas and Metis — no desktop surface at any
of the three.
Decision#
Phase 182 ships a web authoring surface (V1.0) and a mobile surface (V1.1). Desktop is out of scope.
If a desktop surface is ever in scope, the host is apps/yemaya/studio-desktop:
it is the only candidate whose product domain is video production, and it is
three to four times the size of the others. That is a recommendation, not a
commitment.
Rationale#
1. The decisive argument is the biometric copy, not the engineering cost.
The most sensitive artifact in this system is a reference photograph of a real
person. 182.C.24.03 already has to prevent leakage through camera-roll
thumbnails, OS share sheets, logs, clipboard and unprotected caches on mobile. A
desktop host adds a general-purpose filesystem, an unmanaged disk, and a crash
reporter — and 182.C.26 has already established that what leaves our storage
cannot be recalled, only accounted for. A third surface is a third place a
likeness lands, for a capability whose first surface does not exist yet.
2. Choosing a host would settle a repo-wide question this phase has no standing to settle. Two desktop runtimes are live here: Electron in Yemaya, Tauri in Lilith and Iris. Putting the human-video flow on either makes Phase 182 the precedent for a decision that belongs to whoever owns the desktop strategy.
3. Any host would duplicate the whole stack. 182.C.24.05 requires a
desktop surface to use the same server contracts, capability registry, consent,
quality decisions and release proof as web. None of the three candidates has any
of it. The cheapest honest desktop surface is therefore a web view of the web
surface, which is an argument for building the web surface.
4. Nothing is demonstrable yet anywhere. The human-video routes have no caller outside their own specs. A second authoring surface before the first one runs is two surfaces with nothing behind them.
Consequences#
182.C.23targets the web surface (V1.0) and182.C.24the mobile surface (V1.1). Both may now be worked; prohibition 4 is satisfied by this record.182.C.24.05is answered no, with the host named in case the answer changes.- Reversing this needs a named owner, a target release, and an answer to the biometric-copy argument above — not simply a decision to build it.
- The three desktop apps are untouched. Nothing in this ADR deprecates them; they belong to other products.
What this ADR does not decide#
Whether the mobile surface is full-parity authoring or review-focused. That
is 182.C.24.01, and it is a different question: mobile is already on the
release train, so the question there is what belongs on it rather than whether
it exists.