The definitive slot model for the web client, replacing the first-generation
define/register two-step, ScopedSlots whitelist faces, and binding handles:
- 'root' is the only a-priori slot (SlotsService built-in); the shell renders
exactly ctx.slots.renderSlot('root', {}).
- register is the single API: children = slot declaration + render
authorization + runtime spec in one options object; misconfiguration fails
loud at load (duplicate declaration, undeclared contribution, one store
handle under two scopes).
- Component props arrive in four auto-derived shares: PropsRuntime<K>
(owner params + session/global standard kits via declare-merge),
PropsRenderSlots<S>, PropsStore<H>, and the inject business face.
sessionId is framework-supplied; hooks are framework-made only.
- Framework store seat: defineStore factories declare schema/actions/persist;
read = useStore, write = baked actions only; store scope derives from the
mounting entry; per-session persist keys and clearPersisted lifecycle.
- inject factories read the apply closure's own ctx (binding handles retired;
root-ctx back door closed); SessionProvider is self-wired render-prop.
- Rendering sits behind the SlotRenderer install seam; runtime stays
React-free; ownership ledger keyed to the single entry axis closes the
stale-authority window (StaleAuthorizationError probes).
Docs: the slot type-chain note is refreshed in place as the slot system
standard RFC (bilingual pair re-recorded); the web client architecture RFC
defers its slot sections there; packages/client/AGENTS.md gains the slot and
props discipline; gui-testing/web-styling notes drop missions/ references.
Tests: suites rewritten to the standard (props fed directly, real store
engines via createXXXStore().create(), no render machinery); load-time
negative samples for declaration/authorization/store conflicts; verified by
real-host playwright run (three columns, empty state, collapse, keyed session
remount, cross-slot selection sharing).
docs(ui-sidebar): point contract reference at the committed slot standard RFC
missions/ is workspace-local and never committed; the README must not cite it.
@deepseek-ai/dsh-client-ui-slots
Slot registry pure core, slot terminal design: SlotMap declaration merging, the single register composition API on SlotCore, the four-share component-props type family, the store-seat type family, and the renderer install-seam contract. React types only at runtime — the package is React-free and cordis-free.
One register({ name, children?, store?, inject?, ...kind }, Component) call contributes a component into a declared slot and, in the same breath, declares child slots (declaration = render authorization = runtime spec, one table), a store seat, and the registrant's business face. The component is checked at the call site against ComposedProps — the intersection of four shares, each derived from its single source of truth:
| share | type | source |
|---|---|---|
| runtime | PropsRuntime<K> |
SlotMap entry: owner (parent's renderSlot call site) + session standard kit + global seat |
| child render | PropsRenderSlots<S> |
the register call's children key set (statically narrowed renderSlot) |
| store | PropsStore<H> |
the declared handle: useStore selector hook + draft-stripped actions |
| business | I |
inferred from the inject factory's return |
The standard-kit interfaces (SessionStandardProps, GlobalStandardProps) are declared empty here and merged by the runtime package (same declare-merge pattern as SlotMap keys). Inject factory parameters derive from the declaration (InjectParams): session slots get sessionId, a declared store appends baked actions, nothing else — data access lives in the apply closure's ctx.
The store family (defineStore spec in / StoreHandle<T, A> out) types the store seat: init infers the state schema, actions is the complete draft-transform write set, BakedActions strips the draft parameter into the callbacks components and inject factories receive. The defineStore value implementation lives in web-react (the engine's home) and satisfies the DefineStore contract exported here.
SlotCore seeds the a-priori 'root' slot at construction and enforces load-time validation (undeclared-slot registration, duplicate child declaration, one shared handle under two scopes — all throw at register). An entry's disposer collapses its declared child slots recursively: ledger rows, contributions, and store mounts die on one lifecycle axis. renderer.ts carries the install seam (SlotRenderer, SlotRendererHost) plus StaleAuthorizationError/SlotOwnershipError; the implementation lives in web-react, the installation in the shell boot.
Model Experience
None, as the slot registry is browser-side UI plumbing; nothing here reaches a model request.
KV Cache effect
None; this package neither assembles nor sends a provider request.
Known Limitations and Deferred Work
isLivescans all records linearly — fine at UI-plugin registration counts (tens); revisit with an entry→record backref if ledgers ever grow hot.- The
__rendersphantom anchor is visible onPropsRenderSlots— the same accepted noise as the type-chain design's__accepts: generic method signatures compare loosely across key unions, so the contravariant marker is what enforces "component key set ⊆ children declaration".