fix: address ds-review-bot v6/v7 findings on the image-input assembly

- resolveLlmRoute: reuse the yml pi-ai row for providers it already routes
  (DUPLICATE_ADAPTER boot failure) and detect an unset model by origin, not
  by comparison against one deployment default; covered by a new spec.
- LlmService.resolveModelInfoFor preserves (and validates) modality
  metadata, arming the host image preflight for exact-route resolution.
- session.selectModel refuses a text-only target once the session log
  carries an image on any replayed route; an accepted switch would strand
  every later turn with no in-product recovery.
- The composer no longer gates image intake on the handshake activeModel
  snapshot (wrong authority for a per-session decision); the host preflight
  plus the error strip own capability, deployment limits stay client-side.
- InputHub shell teardown releases the scope's draft images (File objects
  and object URLs leaked for the page lifetime).
- session.prompt image parts carry optional alt into the durable block;
  ImageBlock documents assistant-side rendering as forward compatibility.
- Assembled built-client lane apps/web/tests/image-display.snapshot.ts pins
  the history galleries over the authorized attachment route, the lightbox,
  and the composer paste rail; the attachment rail is an accessible group.
- Docs: validateImage on the seam page, fixture byte metadata matches its
  PNG, and the Agent Note claims now match the shipped coverage.

packages/llm/llm/src/index.ts

@@ -333,11 +333,23 @@ export class LlmService extends Service {
         'INVALID_MODEL_CONTEXT',
       )
     }
+    for (const modalities of [resolved.inputModalities, resolved.outputModalities]) {
+      if (modalities !== undefined && (!Array.isArray(modalities) || modalities.some(m => typeof m !== 'string'))) {
+        throw new LlmError(
+          `adapter returned invalid modality metadata for provider "${provider}" model "${model}"`,
+          'INVALID_MODEL_INFO',
+        )
+      }
+    }
     const info: LlmResolvedModelInfo = {
       provider,
       id: model,
       name: resolved.name,
       ...resolved.description === undefined ? {} : { description: resolved.description },
+      // Capability metadata rides through: an explicit modality omission is
+      // negative capability downstream preflights act on (image admission).
+      ...resolved.inputModalities === undefined ? {} : { inputModalities: resolved.inputModalities },
+      ...resolved.outputModalities === undefined ? {} : { outputModalities: resolved.outputModalities },
       ...context === undefined ? {} : { context: { contextWindow: context.contextWindow } },
     }
     const reasoning = resolved.reasoning

packages/llm/llm/src/types.ts

@@ -47,12 +47,17 @@ export interface ReasoningBlock {
   text: string
 }
 
-/** A durable raster image reference, valid in user or assistant content. */
+/**
+ * A durable raster image reference, valid in user or assistant content. The
+ * block is deliberately role-neutral; assistant-side rendering is forward
+ * compatibility — the current production adapters declare text-only output,
+ * so only user content carries images today.
+ */
 export interface ImageBlock {
   type: 'image'
   /** Immutable bytes and intrinsic display metadata owned by the attachment service. */
   attachment: ImageAttachmentRef
-  /** Optional provider- and UI-facing alternative text. */
+  /** Optional provider- and UI-facing alternative text, carried from the prompt wire's image part. */
   alt?: string
 }
 

packages/llm/llm/tests/service.spec.ts

@@ -857,6 +857,8 @@ describe('LlmService', () => {
     [{ provider: 'route', id: 'model', name: 1 }, 'non-string name'],
     [{ provider: 'route', id: 'model', name: '' }, 'empty name'],
     [{ provider: 'route', id: 'model', name: 'Model', description: 1 }, 'non-string description'],
+    [{ provider: 'route', id: 'model', name: 'Model', inputModalities: 'text' }, 'non-array input modalities'],
+    [{ provider: 'route', id: 'model', name: 'Model', outputModalities: [1] }, 'non-string output modality'],
   ] as const)('rejects invalid exact model metadata (%s: %s)', async (metadata, _label) => {
     const ctx = new Context()
     await ctx.plugin(LlmService)
@@ -871,6 +873,27 @@ describe('LlmService', () => {
       .rejects.toMatchObject({ code: 'INVALID_MODEL_INFO' })
   })
 
+  it('preserves modality metadata through exact model resolution', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    const adapter = new class extends ScriptedAdapter {
+      override resolveModel(): Promise<LlmResolvedModelInfo> {
+        return Promise.resolve({
+          provider: 'route', id: 'model', name: 'Model',
+          inputModalities: ['text', 'image'], outputModalities: ['text'],
+        })
+      }
+    }(SCRIPT)
+    ctx.llm.registerAdapter(['route'], adapter)
+
+    // Downstream preflights (image admission) act on this exact field; a
+    // rebuild that drops it silently reads as "modalities unknown".
+    await expect(ctx.llm.resolveModelInfo('route', 'model')).resolves.toEqual({
+      provider: 'route', id: 'model', name: 'Model',
+      inputModalities: ['text', 'image'], outputModalities: ['text'],
+    })
+  })
+
   it('resolves detached model context independently of advisory catalog membership', async () => {
     const ctx = new Context()
     await ctx.plugin(LlmService)