chore: release v0.6.2

Debian docker.io ships no buildx plugin, so the registry-cache

build failed when the runner CLI fell back to it. Install

Docker's official CLI plus buildx plugin instead.

.dockerignore

@@ -1,5 +1,6 @@
 node_modules
 .next
+.worktrees
 .git
 deploy
 docs

.gitea/workflows/release.yml

@@ -31,15 +31,31 @@ jobs:
             echo "image=registry.ginnoir.com/ginnoir/famapp"
           } >> "$GITHUB_OUTPUT"
 
-      - name: Ensure docker CLI
+      - name: Ensure docker CLI + buildx
         run: |
-          if command -v docker >/dev/null 2>&1; then
-            echo "docker already available"
+          # Debian's docker.io package ships no buildx plugin, which the
+          # registry-cache build below requires. Install Docker's official
+          # CLI + buildx plugin so the build works regardless of what the
+          # runner image happens to provide.
+          if docker buildx version >/dev/null 2>&1; then
+            echo "docker + buildx already available"
             docker version
+            docker buildx version
             exit 0
           fi
-          apt-get update -qq && apt-get install -y -qq docker.io
+
+          export DEBIAN_FRONTEND=noninteractive
+          apt-get update -qq
+          apt-get install -y -qq ca-certificates curl gnupg
+          install -m 0755 -d /etc/apt/keyrings
+          curl -fsSL https://download.docker.com/linux/debian/gpg -o /etc/apt/keyrings/docker.asc
+          chmod a+r /etc/apt/keyrings/docker.asc
+          echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/debian $(. /etc/os-release && echo "$VERSION_CODENAME") stable" \
+            > /etc/apt/sources.list.d/docker.list
+          apt-get update -qq
+          apt-get install -y -qq docker-ce-cli docker-buildx-plugin
           docker version
+          docker buildx version
 
       - name: Login to registry
         run: |

CHANGELOG.md

@@ -1,5 +1,31 @@
 # Changelog
 
+## [0.6.2](https://github.com/ginnoir/famapp/compare/v0.6.1...v0.6.2) (2026-07-09)
+
+### Features
+
+- **agent:** add assistant model selector ([4c82d55](https://github.com/ginnoir/famapp/commit/4c82d551ea1f0c8091a3ffc31935a8d22855bbc0))
+- **agent:** add llm model discovery helper ([0bf63cf](https://github.com/ginnoir/famapp/commit/0bf63cf8a83a1adca58fa9e8bf42f16dac60687b))
+- **agent:** expose available assistant models ([27db599](https://github.com/ginnoir/famapp/commit/27db599444fc0bb80d5a1650e2f5b5ba4f695c8e))
+- **agent:** let users choose model route ([ec3f96d](https://github.com/ginnoir/famapp/commit/ec3f96dab170fb2333780438bdec3628cfc58116))
+- **agent:** persist assistant model preference ([bf7e07e](https://github.com/ginnoir/famapp/commit/bf7e07ead91dd8c428f06551afb2113ce838cc82))
+- **agent:** refresh assistant model catalog ([b279f16](https://github.com/ginnoir/famapp/commit/b279f16ba0e1537d2752d30bf5e565f4ede400af))
+- **agent:** route chat through selected model ([9b7a044](https://github.com/ginnoir/famapp/commit/9b7a04431cb12609cbcd766e05fbc411599e0b34))
+
+### Bug Fixes
+
+- **agent:** hide unrelated models for alias fallback ([4dc04b2](https://github.com/ginnoir/famapp/commit/4dc04b21d63aaef7c773f238602f06edbbf03509))
+- **agent:** improve model selector legibility ([bb679d0](https://github.com/ginnoir/famapp/commit/bb679d02be28440acbc860146cbcb3809224aa09))
+- **agent:** move route selector to settings ([49732b0](https://github.com/ginnoir/famapp/commit/49732b02f31cbef57e6cd3b577bee6dbd998e029))
+- **agent:** reject padded assistant model ids ([3983c30](https://github.com/ginnoir/famapp/commit/3983c30fe1b8a324f7d5826df0b24506fe054e50))
+- **agent:** use native model selector ([716bca0](https://github.com/ginnoir/famapp/commit/716bca0fcba8f2026c814f668bf873cf61a0ad16))
+- **agent:** validate assistant model requests ([876e726](https://github.com/ginnoir/famapp/commit/876e72671a0b82b579a9783eb86f452a4a026a52))
+
+### Documentation
+
+- plan assistant model selector ([eb8e562](https://github.com/ginnoir/famapp/commit/eb8e5625656a1e3fe97e19b0bb2514660cf0c5f6))
+- specify assistant model selector ([1a58ef9](https://github.com/ginnoir/famapp/commit/1a58ef993e62ea2855d7d5aafc17446a0e91c33c))
+
 ## [0.6.1](https://github.com/ginnoir/famapp/compare/v0.6.0...v0.6.1) (2026-07-05)
 
 ### Features

docs/superpowers/specs/2026-07-08-assistant-model-selector-design.md

@@ -0,0 +1,151 @@
+# Assistant model selector design
+
+Date: 2026-07-08
+Status: approved for planning
+
+## Context
+
+The AI assistant chat currently uses one environment-configured model through `LLM_MODEL`.
+The chat UI posts messages to `/api/agent/chat`, and the server creates the OpenAI-compatible
+client without any request-time model choice.
+
+ginnoir wants a model selector in the assistant chat. The selector should discover available
+models from the configured OpenAI-compatible provider and save the selected model as the user's
+default.
+
+## Goals
+
+- Show a compact model selector in the assistant chat panel.
+- Discover models from the provider's `/models` endpoint server-side.
+- Persist the selected model per user so it works across browser sessions and devices.
+- Keep `LLM_MODEL` as the fallback when discovery fails, no model is saved, or the saved model
+  is no longer available.
+- Preserve mock-provider behavior in CI and local setups without `LLM_BASE_URL`.
+
+## Non-goals
+
+- Model hosting, training, or fine-tuning.
+- Multiple LLM providers in the same deployment.
+- Per-message experimental settings beyond selecting the model ID.
+- Exposing arbitrary browser-supplied model IDs to the provider.
+
+## User experience
+
+When the assistant bubble opens, the chat panel loads available model IDs from the server.
+The selector appears near the existing assistant status and clear-chat controls. It should be
+visible but compact enough not to reduce the message area materially.
+
+Changing the selector immediately saves the user's default model. The next message uses that
+model, and future assistant sessions start with the saved selection when it is still available.
+
+If model discovery fails, the panel remains usable with the `LLM_MODEL` fallback and shows a
+muted status that model discovery is unavailable. If the saved model has disappeared from the
+provider, the server and UI fall back to `LLM_MODEL`.
+
+## Architecture
+
+### Configuration
+
+`getLlmConfig()` remains the source for provider, base URL, API key, and fallback model.
+No additional allowlist environment variable is required because model IDs come from the
+provider's OpenAI-compatible `/models` endpoint.
+
+### Persistence
+
+Add nullable `assistant_model` storage to `users`.
+
+The existing assistant preference loader should return:
+
+- assistant enabled state
+- assistant display name
+- assistant system prompt
+- saved assistant model ID
+
+The value is nullable. `null` means "use the environment fallback model."
+
+### Model discovery API
+
+Add `GET /api/agent/models`.
+
+Behavior:
+
+- Require the same authenticated user/session or API auth shape as the chat endpoint.
+- Require assistant access to be enabled for the user.
+- If `LLM_BASE_URL` is missing or the provider is mock, return the fallback model as the only
+  available model.
+- Fetch `${LLM_BASE_URL}/models` with `Authorization: Bearer ${LLM_API_KEY}` when configured.
+- Accept OpenAI-style payloads with a top-level `data` array.
+- Normalize each model to `{ id: string, label: string }`, using the ID as the label.
+- Deduplicate, sort consistently, and include the fallback model if the provider omitted it.
+- If discovery fails, return the fallback model plus a degraded status instead of failing the
+  chat UI.
+
+The response should include enough metadata for the UI:
+
+```json
+{
+  "models": [{ "id": "llama3.2", "label": "llama3.2" }],
+  "selectedModel": "llama3.2",
+  "fallbackModel": "llama3.2",
+  "degraded": false
+}
+```
+
+### Saving the default model
+
+Add a server action for updating the user's assistant model, matching the existing assistant
+settings actions. The update path must:
+
+- Accept a model ID string or `null`.
+- Validate length and basic shape before touching the database.
+- Validate the requested model against the current discovered model list.
+- Save `null` when the selected model matches the fallback so `LLM_MODEL` changes take effect for
+  users who have not chosen a non-default model.
+- Revalidate assistant surfaces after saving.
+
+### Chat request flow
+
+Extend `clientChatInputSchema` with optional `model`.
+
+The chat route should:
+
+- Parse `model` from the request body.
+- Resolve the effective model from request model, saved user default, and fallback model.
+- Validate request model and saved user default against discovered models.
+- Reject an invalid request model with `400`.
+- Silently fall back when the saved user default is no longer available.
+- Pass the effective model into `runAgentChat`.
+
+`runAgentChat` should accept an optional model override. `createLlmClient` should support an
+override object or equivalent path that replaces only the model while preserving the configured
+provider, base URL, and API key.
+
+## Error handling
+
+- Missing auth: `401`.
+- Assistant disabled: `403`.
+- Invalid posted model: `400`.
+- Provider `/models` failure: return fallback model from the model-discovery API with
+  `degraded: true`; do not block chat startup.
+- LLM completion failure after a valid model is selected: keep the existing chat error behavior.
+
+## Testing
+
+Unit tests:
+
+- Model discovery normalizes OpenAI-compatible `/models` responses.
+- Discovery falls back to `LLM_MODEL` for mock or failed provider states.
+- Chat input schema accepts an optional valid model string and rejects invalid shapes.
+- Chat route rejects a model not returned by discovery.
+- `runAgentChat` passes the effective model override into the LLM client path.
+
+E2E smoke:
+
+- After assistant opt-in, opening the assistant shows the model selector.
+- Sending a message still renders the user message and assistant response with the mock provider.
+
+## Rollout notes
+
+This is additive. Existing deployments without a provider `/models` endpoint continue to use
+`LLM_MODEL`. The database migration is nullable, so existing users keep current behavior until
+they choose a model.

drizzle/0025_assistant_model.sql

@@ -0,0 +1 @@
+ALTER TABLE "users" ADD COLUMN "assistant_model" text;

drizzle/0026_assistant_model_route.sql

@@ -0,0 +1,5 @@
+ALTER TABLE "users" ADD COLUMN "assistant_model_route" text;--> statement-breakpoint
+UPDATE "users"
+SET "assistant_model_route" = "assistant_model",
+    "assistant_model" = NULL
+WHERE "assistant_model" IN ('auto', 'uncensored');

drizzle/meta/_journal.json

@@ -176,6 +176,20 @@
       "when": 1780394000000,
       "tag": "0024_user_assistant_customization",
       "breakpoints": true
+    },
+    {
+      "idx": 25,
+      "version": "7",
+      "when": 1783560000000,
+      "tag": "0025_assistant_model",
+      "breakpoints": true
+    },
+    {
+      "idx": 26,
+      "version": "7",
+      "when": 1783561000000,
+      "tag": "0026_assistant_model_route",
+      "breakpoints": true
     }
   ]
 }

eslint.config.mjs

@@ -8,6 +8,7 @@ export default tseslint.config(
     ignores: [
       "node_modules/**",
       ".next/**",
+      ".worktrees/**",
       ".claude/**",
       ".design-tmp/**",
       "dist/**",

next.config.ts

@@ -178,7 +178,14 @@ const nextConfig: NextConfig = {
   output: "standalone",
   // Keep pino and pino-pretty as native Node.js requires so their worker-thread
   // transport and stream internals work correctly inside the standalone bundle.
-  serverExternalPackages: ["pino", "pino-pretty", "drizzle-orm", "postgres"],
+  serverExternalPackages: [
+    "pino",
+    "pino-pretty",
+    "drizzle-orm",
+    "postgres",
+    "isomorphic-dompurify",
+    "jsdom",
+  ],
 };
 
 export default nextConfig;

package.json

@@ -1,6 +1,6 @@
 {
   "name": "famapp",
-  "version": "0.6.1",
+  "version": "0.6.2",
   "private": true,
   "type": "module",
   "packageManager": "pnpm@10.33.3",

src/app/api/agent/chat/route.ts

@@ -2,6 +2,7 @@ import { apiError, apiJson } from "@/lib/api-handler";
 import { resolveApiAuth } from "@/lib/api-auth";
 import { getAssistantPreferences, resolveAssistantSystemPrompt } from "@/lib/assistant-preference";
 import { isLlmConfigured } from "@/lib/llm";
+import { listLlmModels, resolveAssistantModel } from "@/lib/llm/models";
 import { clientChatInputSchema } from "@/modules/agent/messages";
 import { encodeSseEvent } from "@/modules/agent/server/progress";
 import { runAgentChat } from "@/modules/agent/server/run";
@@ -31,6 +32,18 @@ export async function POST(request: Request) {
     return apiError(parsed.error.issues[0]?.message ?? "Validation error", 400);
   }
 
+  const modelList = await listLlmModels({ route: assistant.modelRoute });
+  const modelResolution = resolveAssistantModel({
+    requestedModel: parsed.data.model,
+    savedModel: assistant.model,
+    fallbackModel: modelList.fallbackModel,
+    models: modelList.models,
+  });
+
+  if (!modelResolution.ok) {
+    return apiError(modelResolution.error, 400);
+  }
+
   if (parsed.data.stream) {
     const stream = new ReadableStream<Uint8Array>({
       async start(controller) {
@@ -44,6 +57,7 @@ export async function POST(request: Request) {
             messages: parsed.data.messages,
             request,
             systemPrompt,
+            model: modelResolution.model,
             onProgress: send,
           });
 
@@ -78,6 +92,7 @@ export async function POST(request: Request) {
       messages: parsed.data.messages,
       request,
       systemPrompt,
+      model: modelResolution.model,
     });
 
     return apiJson({

src/app/api/agent/models/route.ts

@@ -0,0 +1,47 @@
+import { apiError, apiJson } from "@/lib/api-handler";
+import { resolveApiAuth } from "@/lib/api-auth";
+import { getAssistantPreferences } from "@/lib/assistant-preference";
+import { isValidAssistantModelRoute, listLlmModels, resolveAssistantModel } from "@/lib/llm/models";
+
+export const dynamic = "force-dynamic";
+
+function noStore(response: Response): Response {
+  response.headers.set("Cache-Control", "no-store");
+  return response;
+}
+
+export async function GET(request: Request) {
+  const auth = await resolveApiAuth(request);
+  if (!auth?.userId) {
+    return noStore(apiError("Unauthorized", 401));
+  }
+
+  const assistant = await getAssistantPreferences(auth.userId);
+  if (!assistant.enabled) {
+    return noStore(apiError("Assistant not enabled", 403));
+  }
+
+  const requestedRoute = new URL(request.url).searchParams.get("route");
+  if (requestedRoute !== null && !isValidAssistantModelRoute(requestedRoute)) {
+    return noStore(apiError("Invalid assistant model route", 400));
+  }
+
+  const modelRoute = requestedRoute ?? assistant.modelRoute;
+  const modelList = await listLlmModels({ route: modelRoute });
+  const resolved = resolveAssistantModel({
+    requestedModel: null,
+    savedModel: requestedRoute === null ? assistant.model : null,
+    fallbackModel: modelList.fallbackModel,
+    models: modelList.models,
+  });
+
+  return noStore(
+    apiJson({
+      models: modelList.models,
+      selectedModel: resolved.model,
+      fallbackModel: modelList.fallbackModel,
+      route: modelList.route,
+      degraded: modelList.degraded,
+    }),
+  );
+}

src/app/layout.tsx

@@ -104,6 +104,7 @@ export default async function RootLayout({ children }: { children: React.ReactNo
   let signedIn = false;
   let assistantEnabled = false;
   let assistantName = DEFAULT_ASSISTANT_NAME;
+  let assistantModel: string | null = null;
 
   const session = await auth();
   if (session?.user?.id) {
@@ -117,6 +118,7 @@ export default async function RootLayout({ children }: { children: React.ReactNo
         themeNavStyle: users.themeNavStyle,
         assistantEnabled: users.assistantEnabled,
         assistantName: users.assistantName,
+        assistantModel: users.assistantModel,
       })
       .from(users)
       .where(eq(users.id, session.user.id))
@@ -129,6 +131,7 @@ export default async function RootLayout({ children }: { children: React.ReactNo
       navStyle = row.themeNavStyle as NavStyle;
       assistantEnabled = row.assistantEnabled;
       assistantName = row.assistantName?.trim() || DEFAULT_ASSISTANT_NAME;
+      assistantModel = row.assistantModel?.trim() || null;
     }
     userDashboards = await db
       .select({
@@ -186,6 +189,7 @@ export default async function RootLayout({ children }: { children: React.ReactNo
               configured={isLlmConfigured()}
               userId={session.user.id}
               assistantName={assistantName}
+              assistantModel={assistantModel}
             />
           ) : null}
           <AppToaster position="bottom-right" />

src/app/settings/assistant-actions.ts

@@ -9,8 +9,15 @@ import {
   MAX_ASSISTANT_NAME_LENGTH,
   MAX_ASSISTANT_SYSTEM_PROMPT_LENGTH,
 } from "@/lib/assistant-config";
+import {
+  isValidAssistantModelRoute,
+  isValidLlmModelId,
+  listLlmModels,
+  type AssistantModelRoute,
+} from "@/lib/llm/models";
 import { users } from "@/modules/_core/schema";
 import { getCurrentSession } from "@/lib/session";
+import { getAssistantPreferences } from "@/lib/assistant-preference";
 
 const assistantNameSchema = z
   .string()
@@ -50,6 +57,39 @@ export async function setAssistantSystemPrompt(prompt: string | null): Promise<v
   revalidateAssistantSurfaces();
 }
 
+export async function setAssistantModelRoute(route: AssistantModelRoute): Promise<void> {
+  if (!isValidAssistantModelRoute(route)) {
+    throw new Error("Invalid assistant model route");
+  }
+
+  const { user } = await getCurrentSession();
+  await db
+    .update(users)
+    .set({ assistantModelRoute: route, assistantModel: null })
+    .where(eq(users.id, user.id));
+  revalidateAssistantSurfaces();
+}
+
+export async function setAssistantModel(model: string | null): Promise<void> {
+  const { user } = await getCurrentSession();
+  const normalized = model?.trim() || null;
+
+  if (normalized !== null && !isValidLlmModelId(normalized)) {
+    throw new Error("Invalid assistant model");
+  }
+
+  const assistant = await getAssistantPreferences(user.id);
+  const available = await listLlmModels({ route: assistant.modelRoute });
+  const requested = normalized === available.fallbackModel ? null : normalized;
+
+  if (requested !== null && !available.models.some((option) => option.id === requested)) {
+    throw new Error("Invalid assistant model");
+  }
+
+  await db.update(users).set({ assistantModel: requested }).where(eq(users.id, user.id));
+  revalidateAssistantSurfaces();
+}
+
 export async function resetAssistantName(): Promise<void> {
   await setAssistantName(DEFAULT_ASSISTANT_NAME);
 }

src/app/settings/page.tsx

@@ -324,6 +324,7 @@ function AppearanceSection({
     themeNavStyle: string;
     assistantEnabled: boolean;
     assistantName: string;
+    assistantModelRoute: string | null;
     assistantSystemPrompt: string | null;
   };
 }) {
@@ -361,6 +362,7 @@ function AppearanceSection({
           <AssistantSettings
             enabled={user.assistantEnabled}
             name={user.assistantName}
+            modelRoute={user.assistantModelRoute}
             systemPrompt={user.assistantSystemPrompt}
             defaultSystemPrompt={AGENT_SYSTEM_PROMPT}
           />

src/components/assistant-settings.tsx

@@ -7,9 +7,11 @@ import {
   resetAssistantSystemPrompt,
   setAssistantEnabled,
   setAssistantName,
+  setAssistantModelRoute,
   setAssistantSystemPrompt,
 } from "@/app/settings/assistant-actions";
 import { DEFAULT_ASSISTANT_NAME, MAX_ASSISTANT_NAME_LENGTH } from "@/lib/assistant-config";
+import type { AssistantModelRoute } from "@/lib/llm/models";
 import { Button } from "@/components/ui/button";
 import { Input } from "@/components/ui/input";
 import { Switch } from "@/components/ui/switch";
@@ -17,15 +19,29 @@ import { Switch } from "@/components/ui/switch";
 type Props = {
   enabled: boolean;
   name: string;
+  modelRoute: string | null;
   systemPrompt: string | null;
   defaultSystemPrompt: string;
 };
 
-export function AssistantSettings({ enabled, name, systemPrompt, defaultSystemPrompt }: Props) {
+function normalizeAssistantModelRoute(value: string | null): AssistantModelRoute {
+  return value === "uncensored" ? "uncensored" : "auto";
+}
+
+export function AssistantSettings({
+  enabled,
+  name,
+  modelRoute,
+  systemPrompt,
+  defaultSystemPrompt,
+}: Props) {
   const [isPending, startTransition] = useTransition();
   const router = useRouter();
 
   const effectivePrompt = systemPrompt ?? defaultSystemPrompt;
+  const [selectedRoute, setSelectedRoute] = useState<AssistantModelRoute>(() =>
+    normalizeAssistantModelRoute(modelRoute),
+  );
   const [savedName, setSavedName] = useState(name);
   const [draftName, setDraftName] = useState(name);
   const [savedPrompt, setSavedPrompt] = useState(systemPrompt);
@@ -45,6 +61,16 @@ export function AssistantSettings({ enabled, name, systemPrompt, defaultSystemPr
     });
   }
 
+  function changeModelRoute(nextRoute: string) {
+    const route = normalizeAssistantModelRoute(nextRoute);
+    setSelectedRoute(route);
+
+    startTransition(async () => {
+      await setAssistantModelRoute(route);
+      router.refresh();
+    });
+  }
+
   function saveName() {
     const next = draftName.trim();
     if (!next) return;
@@ -105,6 +131,29 @@ export function AssistantSettings({ enabled, name, systemPrompt, defaultSystemPr
         />
       </label>
 
+      <div className="flex flex-col gap-2">
+        <div className="min-w-0">
+          <label htmlFor="assistant-model-route" className="text-sm font-medium">
+            Model route
+          </label>
+          <p className="muted text-[12px] mt-0.5">
+            Choose the default router family for your assistant. Individual models can still be
+            picked from the chat panel.
+          </p>
+        </div>
+        <select
+          id="assistant-model-route"
+          aria-label="Assistant model route"
+          value={selectedRoute}
+          disabled={isPending}
+          onChange={(event) => changeModelRoute(event.target.value)}
+          className="input h-10"
+        >
+          <option value="auto">Auto</option>
+          <option value="uncensored">Uncensored</option>
+        </select>
+      </div>
+
       <div className="flex flex-col gap-2">
         <div className="flex items-end justify-between gap-3">
           <div className="min-w-0 flex-1">

src/lib/assistant-preference.ts

@@ -1,6 +1,7 @@
 import { eq } from "drizzle-orm";
 import { db } from "@/lib/db";
 import { DEFAULT_ASSISTANT_NAME } from "@/lib/assistant-config";
+import { isValidAssistantModelRoute, type AssistantModelRoute } from "@/lib/llm/models";
 import { users } from "@/modules/_core/schema";
 
 export {
@@ -14,6 +15,8 @@ export type AssistantPreferences = {
   enabled: boolean;
   name: string;
   systemPrompt: string | null;
+  modelRoute: AssistantModelRoute | null;
+  model: string | null;
 };
 
 export async function getAssistantPreferences(userId: string): Promise<AssistantPreferences> {
@@ -22,6 +25,8 @@ export async function getAssistantPreferences(userId: string): Promise<Assistant
       assistantEnabled: users.assistantEnabled,
       assistantName: users.assistantName,
       assistantSystemPrompt: users.assistantSystemPrompt,
+      assistantModelRoute: users.assistantModelRoute,
+      assistantModel: users.assistantModel,
     })
     .from(users)
     .where(eq(users.id, userId))
@@ -31,6 +36,11 @@ export async function getAssistantPreferences(userId: string): Promise<Assistant
     enabled: row?.assistantEnabled ?? false,
     name: row?.assistantName?.trim() || DEFAULT_ASSISTANT_NAME,
     systemPrompt: row?.assistantSystemPrompt ?? null,
+    modelRoute:
+      row?.assistantModelRoute && isValidAssistantModelRoute(row.assistantModelRoute)
+        ? row.assistantModelRoute
+        : null,
+    model: row?.assistantModel?.trim() || null,
   };
 }
 

src/lib/llm/index.ts

@@ -13,8 +13,8 @@ export type {
 export { getLlmConfig, isLlmConfigured } from "./config";
 export { createMockLlmClient } from "./mock";
 
-export function createLlmClient(override?: LlmClient): LlmClient {
-  if (override) return override;
+export function createLlmClient(options?: { model?: string; override?: LlmClient }): LlmClient {
+  if (options?.override) return options.override;
 
   const config = getLlmConfig();
   if (config.provider === "mock" || !config.baseUrl) {
@@ -24,6 +24,6 @@ export function createLlmClient(override?: LlmClient): LlmClient {
   return createOpenAiCompatibleClient({
     baseUrl: config.baseUrl,
     apiKey: config.apiKey,
-    model: config.model,
+    model: options?.model ?? config.model,
   });
 }

src/lib/llm/models.ts

@@ -0,0 +1,148 @@
+import { getLlmConfig, type LlmConfig } from "./config";
+
+export type LlmModelOption = {
+  id: string;
+  label: string;
+};
+
+export const ASSISTANT_MODEL_ROUTES = ["auto", "uncensored"] as const;
+
+export type AssistantModelRoute = (typeof ASSISTANT_MODEL_ROUTES)[number];
+
+export type LlmModelsResult = {
+  models: LlmModelOption[];
+  fallbackModel: string;
+  route: AssistantModelRoute;
+  degraded: boolean;
+};
+
+export type AssistantModelResolution =
+  | { ok: true; model: string }
+  | { ok: false; model: string; error: string };
+
+const MODEL_ID_PATTERN = /^[A-Za-z0-9._:/-]+$/;
+const MAX_MODEL_ID_LENGTH = 128;
+
+export function isValidLlmModelId(value: string): boolean {
+  const trimmed = value.trim();
+  return (
+    trimmed.length > 0 &&
+    trimmed.length <= MAX_MODEL_ID_LENGTH &&
+    trimmed === value &&
+    MODEL_ID_PATTERN.test(trimmed)
+  );
+}
+
+export function isValidAssistantModelRoute(value: string): value is AssistantModelRoute {
+  return ASSISTANT_MODEL_ROUTES.some((route) => route === value);
+}
+
+function resolveModelRoute(
+  route: AssistantModelRoute | null | undefined,
+  config: LlmConfig,
+): { route: AssistantModelRoute; fallbackModel: string } {
+  if (route) {
+    return { route, fallbackModel: route };
+  }
+
+  const defaultRoute: AssistantModelRoute = config.model === "uncensored" ? "uncensored" : "auto";
+  return { route: defaultRoute, fallbackModel: config.model };
+}
+
+export function normalizeLlmModelsPayload(payload: unknown): LlmModelOption[] {
+  const data =
+    typeof payload === "object" && payload !== null && "data" in payload
+      ? (payload as { data?: unknown }).data
+      : null;
+
+  if (!Array.isArray(data)) return [];
+
+  const ids = new Set<string>();
+  for (const row of data) {
+    if (typeof row !== "object" || row === null || !("id" in row)) continue;
+    const id = (row as { id?: unknown }).id;
+    if (typeof id !== "string") continue;
+    if (!isValidLlmModelId(id)) continue;
+    ids.add(id);
+  }
+
+  return [...ids].sort((a, b) => a.localeCompare(b)).map((id) => ({ id, label: id }));
+}
+
+export async function listLlmModels(options?: {
+  config?: LlmConfig;
+  fetchImpl?: typeof fetch;
+  route?: AssistantModelRoute | null;
+}): Promise<LlmModelsResult> {
+  const config = options?.config ?? getLlmConfig();
+  const fetchImpl = options?.fetchImpl ?? fetch;
+  const { route, fallbackModel } = resolveModelRoute(options?.route, config);
+  const fallbackOption = { id: fallbackModel, label: fallbackModel };
+
+  if (config.provider === "mock" || !config.baseUrl) {
+    return { models: [fallbackOption], fallbackModel, route, degraded: false };
+  }
+
+  try {
+    const headers: Record<string, string> = {};
+    if (config.apiKey) headers.Authorization = `Bearer ${config.apiKey}`;
+
+    const modelsUrl = new URL(`${config.baseUrl.replace(/\/$/, "")}/models`);
+    if (route === "uncensored") {
+      modelsUrl.searchParams.set("type", "uncensored");
+    }
+
+    const response = await fetchImpl(modelsUrl, {
+      method: "GET",
+      headers,
+    });
+
+    if (!response.ok) {
+      return { models: [fallbackOption], fallbackModel, route, degraded: true };
+    }
+
+    const models = normalizeLlmModelsPayload(await response.json());
+
+    if (models.length === 0) {
+      return { models: [fallbackOption], fallbackModel, route, degraded: true };
+    }
+
+    if (!models.some((model) => model.id === fallbackModel)) {
+      return { models: [fallbackOption], fallbackModel, route, degraded: false };
+    }
+
+    return {
+      models,
+      fallbackModel,
+      route,
+      degraded: false,
+    };
+  } catch {
+    return { models: [fallbackOption], fallbackModel, route, degraded: true };
+  }
+}
+
+export function resolveAssistantModel(options: {
+  requestedModel: string | null | undefined;
+  savedModel: string | null | undefined;
+  fallbackModel: string;
+  models: LlmModelOption[];
+}): AssistantModelResolution {
+  const available = new Set(options.models.map((model) => model.id));
+  const fallback = available.has(options.fallbackModel)
+    ? options.fallbackModel
+    : (options.models[0]?.id ?? options.fallbackModel);
+
+  if (options.requestedModel !== null && options.requestedModel !== undefined) {
+    if (!isValidLlmModelId(options.requestedModel) || !available.has(options.requestedModel)) {
+      return { ok: false, model: fallback, error: "Invalid assistant model" };
+    }
+    return { ok: true, model: options.requestedModel };
+  }
+
+  if (options.savedModel && available.has(options.savedModel)) {
+    return { ok: true, model: options.savedModel };
+  }
+
+  return { ok: true, model: fallback };
+}

src/modules/_core/schema.ts

@@ -38,6 +38,8 @@ export const users = pgTable("users", {
   assistantEnabled: boolean("assistant_enabled").notNull().default(false),
   assistantName: text("assistant_name").notNull().default("Assistant"),
   assistantSystemPrompt: text("assistant_system_prompt"),
+  assistantModelRoute: text("assistant_model_route"),
+  assistantModel: text("assistant_model"),
   defaultEventReminderOffsets: jsonb("default_event_reminder_offsets")
     .notNull()
     .$type<number[]>()

src/modules/agent/components/assistant-bubble.tsx

@@ -8,9 +8,10 @@ type Props = {
   configured: boolean;
   userId: string;
   assistantName: string;
+  assistantModel: string | null;
 };
 
-export function AssistantBubble({ configured, userId, assistantName }: Props) {
+export function AssistantBubble({ configured, userId, assistantName, assistantModel }: Props) {
   const [open, setOpen] = useState(false);
 
   return (
@@ -41,6 +42,7 @@ export function AssistantBubble({ configured, userId, assistantName }: Props) {
             configured={configured}
             userId={userId}
             assistantName={assistantName}
+            assistantModel={assistantModel}
           />
         </div>
       ) : null}

src/modules/agent/components/assistant-panel.tsx

@@ -1,7 +1,8 @@
 "use client";
 
-import { useEffect, useRef, useState } from "react";
-import { ImagePlus, Loader2, Mic, Send, Square } from "lucide-react";
+import { useCallback, useEffect, useRef, useState, useTransition } from "react";
+import { ChevronDown, ImagePlus, Loader2, Mic, RefreshCw, Send, Square } from "lucide-react";
+import { setAssistantModel } from "@/app/settings/assistant-actions";
 import { Button } from "@/components/ui/button";
 import { Input } from "@/components/ui/input";
 import { consumeAgentChatStream } from "../assistant-chat-stream";
@@ -18,12 +19,25 @@ type Props = {
   configured: boolean;
   userId: string;
   assistantName: string;
+  assistantModel: string | null;
 };
 
 type PendingImage = {
   url: string;
 };
 
+type LlmModelOption = {
+  id: string;
+  label: string;
+};
+
+type ModelsResponse = {
+  models: LlmModelOption[];
+  selectedModel: string;
+  fallbackModel: string;
+  degraded: boolean;
+};
+
 async function uploadAssistantImage(file: File): Promise<string> {
   const formData = new FormData();
   formData.append("file", file);
@@ -36,7 +50,7 @@ async function uploadAssistantImage(file: File): Promise<string> {
   return payload.url;
 }
 
-export function AssistantPanel({ configured, userId, assistantName }: Props) {
+export function AssistantPanel({ configured, userId, assistantName, assistantModel }: Props) {
   const [messages, setMessages] = useState<AssistantChatMessage[]>(() => loadAssistantChat(userId));
   const [input, setInput] = useState("");
   const [pendingImage, setPendingImage] = useState<PendingImage | null>(null);
@@ -44,6 +58,12 @@ export function AssistantPanel({ configured, userId, assistantName }: Props) {
   const [error, setError] = useState<string | null>(null);
   const [isPending, setIsPending] = useState(false);
   const [activityLabel, setActivityLabel] = useState<string | null>(null);
+  const [modelOptions, setModelOptions] = useState<LlmModelOption[]>([]);
+  const [selectedModel, setSelectedModel] = useState(assistantModel ?? "");
+  const [fallbackModel, setFallbackModel] = useState("");
+  const [modelsDegraded, setModelsDegraded] = useState(false);
+  const [modelsLoading, setModelsLoading] = useState(true);
+  const [savingModel, startTransition] = useTransition();
   const listRef = useRef<HTMLDivElement>(null);
   const abortRef = useRef<AbortController | null>(null);
   const imageInputRef = useRef<HTMLInputElement>(null);
@@ -67,6 +87,48 @@ export function AssistantPanel({ configured, userId, assistantName }: Props) {
     };
   }, []);
 
+  const loadModels = useCallback(async (options?: { refresh?: boolean; signal?: AbortSignal }) => {
+    if (options?.signal?.aborted) return;
+
+    setModelsLoading(true);
+    try {
+      const params = new URLSearchParams();
+      if (options?.refresh) params.set("refresh", "1");
+      const query = params.size > 0 ? `?${params.toString()}` : "";
+
+      const response = await fetch(`/api/agent/models${query}`, {
+        cache: "no-store",
+        signal: options?.signal,
+      });
+      if (!response.ok) throw new Error("Model discovery unavailable");
+      const payload = (await response.json()) as ModelsResponse;
+      if (options?.signal?.aborted) return;
+      setModelOptions(payload.models);
+      setSelectedModel(payload.selectedModel);
+      setFallbackModel(payload.fallbackModel);
+      setModelsDegraded(payload.degraded);
+      setError(null);
+    } catch (err) {
+      if (err instanceof Error && err.name === "AbortError") return;
+      setModelsDegraded(true);
+      setError("Model discovery unavailable");
+    } finally {
+      if (!options?.signal?.aborted) setModelsLoading(false);
+    }
+  }, []);
+
+  useEffect(() => {
+    const controller = new AbortController();
+
+    queueMicrotask(() => {
+      void loadModels({ signal: controller.signal });
+    });
+
+    return () => {
+      controller.abort();
+    };
+  }, [loadModels]);
+
   function scrollToBottom() {
     requestAnimationFrame(() => {
       const node = listRef.current;
@@ -102,6 +164,21 @@ export function AssistantPanel({ configured, userId, assistantName }: Props) {
     }
   }
 
+  function changeModel(nextModel: string | null) {
+    if (!nextModel) return;
+
+    setSelectedModel(nextModel);
+    setError(null);
+
+    startTransition(async () => {
+      try {
+        await setAssistantModel(nextModel === fallbackModel ? null : nextModel);
+      } catch (err) {
+        setError(err instanceof Error ? err.message : "Could not save assistant model");
+      }
+    });
+  }
+
   async function sendMessage() {
     const text = input.trim();
     const hasImage = pendingImage !== null;
@@ -133,6 +210,7 @@ export function AssistantPanel({ configured, userId, assistantName }: Props) {
         headers: { "Content-Type": "application/json" },
         body: JSON.stringify({
           messages: nextMessages.map(toClientChatMessage),
+          model: selectedModel || undefined,
           stream: true,
         }),
         signal: controller.signal,
@@ -168,6 +246,7 @@ export function AssistantPanel({ configured, userId, assistantName }: Props) {
   const inputDisabled = isPending || voiceState === "transcribing" || uploadingImage;
   const canSend =
     !isPending &&
+    !savingModel &&
     voiceState === "idle" &&
     !uploadingImage &&
     (input.trim().length > 0 || pendingImage !== null);
@@ -175,21 +254,63 @@ export function AssistantPanel({ configured, userId, assistantName }: Props) {
   return (
     <div className="flex min-h-0 flex-1 flex-col gap-3">
       <div className="flex items-start justify-between gap-3">
-        <p className="muted min-w-0 text-[12px] leading-relaxed">
-          {configured
-            ? "Type, talk, or send a photo — I can update lists, calendar, notes, and more."
-            : "Mock provider active — set LLM_BASE_URL for your homelab model."}
-        </p>
-        {messages.length > 0 ? (
-          <button
+        <div className="min-w-0 flex-1">
+          <p className="muted text-[12px] leading-relaxed">
+            {configured
+              ? "Type, talk, or send a photo — I can update lists, calendar, notes, and more."
+              : "Mock provider active — set LLM_BASE_URL for your homelab model."}
+          </p>
+          {modelsDegraded ? (
+            <p className="muted mt-1 text-[11px]">Model discovery unavailable; using fallback.</p>
+          ) : null}
+        </div>
+        <div className="flex shrink-0 items-center gap-2">
+          {modelOptions.length > 0 ? (
+            <div className="relative max-w-36">
+              <select
+                aria-label="Assistant model"
+                value={selectedModel}
+                onChange={(event) => changeModel(event.target.value)}
+                disabled={modelsLoading || savingModel || isPending}
+                className="h-9 w-full max-w-36 appearance-none truncate rounded-[min(var(--radius-md),10px)] border border-input bg-[var(--card)] py-0 pr-8 pl-3 text-[13px] leading-9 text-[var(--ink)] outline-none transition-colors focus-visible:border-ring focus-visible:ring-3 focus-visible:ring-ring/50 disabled:cursor-not-allowed disabled:opacity-50"
+              >
+                {modelOptions.map((model) => (
+                  <option key={model.id} value={model.id}>
+                    {model.label}
+                  </option>
+                ))}
+              </select>
+              <ChevronDown
+                className="pointer-events-none absolute top-1/2 right-2 size-4 -translate-y-1/2 text-muted-foreground"
+                aria-hidden="true"
+              />
+            </div>
+          ) : null}
+          <Button
             type="button"
-            onClick={clearChat}
-            disabled={isPending}
-            className="shrink-0 text-[11px] text-muted-foreground transition-colors hover:text-foreground disabled:opacity-50"
+            size="sm"
+            variant="outline"
+            aria-label="Refresh assistant models"
+            disabled={modelsLoading || savingModel || isPending}
+            onClick={() => void loadModels({ refresh: true })}
           >
-            Clear
-          </button>
-        ) : null}
+            {modelsLoading ? (
+              <Loader2 className="size-4 animate-spin" />
+            ) : (
+              <RefreshCw className="size-4" />
+            )}
+          </Button>
+          {messages.length > 0 ? (
+            <button
+              type="button"
+              onClick={clearChat}
+              disabled={isPending}
+              className="shrink-0 text-[11px] text-muted-foreground transition-colors hover:text-foreground disabled:opacity-50"
+            >
+              Clear
+            </button>
+          ) : null}
+        </div>
       </div>
 
       <div

src/modules/agent/messages.ts

@@ -1,4 +1,5 @@
 import { z } from "zod";
+import { isValidLlmModelId } from "@/lib/llm/models";
 
 export const clientChatAttachmentSchema = z.object({
   type: z.literal("image"),
@@ -11,8 +12,13 @@ export const clientChatMessageSchema = z.object({
   attachments: z.array(clientChatAttachmentSchema).max(3).optional(),
 });
 
+export const clientChatModelSchema = z
+  .string()
+  .refine((value) => isValidLlmModelId(value), "Invalid assistant model");
+
 export const clientChatInputSchema = z.object({
   stream: z.boolean().optional(),
+  model: clientChatModelSchema.optional(),
   messages: z.array(clientChatMessageSchema).min(1).max(40),
 });
 

src/modules/agent/server/run.ts

@@ -45,11 +45,12 @@ export async function runAgentChat(options: {
   messages: ClientChatMessage[];
   request: Request;
   systemPrompt?: string;
+  model?: string;
   llm?: LlmClient;
   executeTool?: ToolExecutor;
   onProgress?: AgentProgressHandler;
 }): Promise<AgentChatResult> {
-  const llm = options.llm ?? createLlmClient();
+  const llm = options.llm ?? createLlmClient({ model: options.model });
   const executeTool = options.executeTool ?? createApiToolExecutor(options.request);
   const onProgress = options.onProgress;
   const systemPrompt = options.systemPrompt ?? AGENT_SYSTEM_PROMPT;

tests/e2e/assistant.spec.ts

@@ -1,4 +1,13 @@
-import { expect, test } from "@playwright/test";
+import { expect, test, type Page } from "@playwright/test";
+
+async function ensureSignedIn(page: Page) {
+  await page.goto("/");
+  const devLogin = page.getByRole("button", { name: "Dev login" });
+  if (await devLogin.isVisible().catch(() => false)) {
+    await devLogin.click();
+    await page.waitForURL((url) => !url.pathname.startsWith("/login"));
+  }
+}
 
 test("assistant bubble is hidden until opted in", async ({ page }) => {
   await page.goto("/");
@@ -6,15 +15,35 @@ test("assistant bubble is hidden until opted in", async ({ page }) => {
 });
 
 test("assistant chat smoke after opt-in", async ({ page }) => {
+  await ensureSignedIn(page);
   await page.goto("/settings?s=appearance");
   const assistantSwitch = page.getByRole("switch", { name: "AI assistant" });
   if (!(await assistantSwitch.isChecked())) {
     await assistantSwitch.click();
   }
+  await expect(assistantSwitch).toBeChecked();
+  const routeSelector = page.getByRole("combobox", { name: "Assistant model route" });
+  await expect(routeSelector).toBeVisible();
+  await routeSelector.selectOption("auto");
+  await expect(routeSelector).toHaveValue("auto");
+  await expect(routeSelector).toBeEnabled();
+  await routeSelector.selectOption("uncensored");
+  await expect(routeSelector).toHaveValue("uncensored");
+  await expect(routeSelector).toBeEnabled();
 
   await page.goto("/");
   await page.getByRole("button", { name: "Open assistant" }).click();
   await expect(page.getByRole("dialog", { name: "Assistant" })).toBeVisible();
+  await expect(page.getByRole("combobox", { name: "Assistant model route" })).toHaveCount(0);
+  const modelSelector = page.getByRole("combobox", { name: "Assistant model", exact: true });
+  await expect(modelSelector).toBeVisible();
+  const refreshModels = page.getByRole("button", { name: "Refresh assistant models" });
+  await expect(refreshModels).toBeVisible();
+  await refreshModels.click();
+  await expect.poll(() => modelSelector.evaluate((node) => node.tagName)).toBe("SELECT");
+  await expect
+    .poll(() => modelSelector.evaluate((node) => node.getBoundingClientRect().height))
+    .toBeGreaterThanOrEqual(36);
 
   await page.getByLabel("Message for Assistant").fill("hello assistant");
   await page.getByRole("button", { name: "Send" }).click();

tests/unit/agent-chat.test.ts

@@ -84,3 +84,36 @@ describe("runAgentChat", () => {
     assert.ok(result.message.content.length > 0);
   });
 });
+
+it("passes a model override to the OpenAI-compatible client", async () => {
+  const originalBaseUrl = process.env.LLM_BASE_URL;
+  const originalModel = process.env.LLM_MODEL;
+  const originalProvider = process.env.LLM_PROVIDER;
+  const originalFetch = globalThis.fetch;
+  let requestBody: unknown = null;
+
+  process.env.LLM_BASE_URL = "https://llm.example.test/v1";
+  process.env.LLM_MODEL = "llama3.2";
+  delete process.env.LLM_PROVIDER;
+
+  globalThis.fetch = (async (_input: RequestInfo | URL, init?: RequestInit) => {
+    requestBody = JSON.parse(String(init?.body));
+    return Response.json({
+      choices: [{ message: { role: "assistant", content: "done" }, finish_reason: "stop" }],
+    });
+  }) as typeof fetch;
+
+  const { createLlmClient } = await import("../../src/lib/llm/index");
+  const client = createLlmClient({ model: "qwen2.5-coder" });
+  await client.chatCompletion({ messages: [{ role: "user", content: "hello" }] });
+
+  assert.equal((requestBody as { model?: string }).model, "qwen2.5-coder");
+
+  globalThis.fetch = originalFetch;
+  if (originalBaseUrl === undefined) delete process.env.LLM_BASE_URL;
+  else process.env.LLM_BASE_URL = originalBaseUrl;
+  if (originalModel === undefined) delete process.env.LLM_MODEL;
+  else process.env.LLM_MODEL = originalModel;
+  if (originalProvider === undefined) delete process.env.LLM_PROVIDER;
+  else process.env.LLM_PROVIDER = originalProvider;
+});

tests/unit/agent-messages.test.ts

@@ -25,4 +25,31 @@ describe("clientChatInputSchema", () => {
 
     assert.equal(parsed.success, false);
   });
+
+  it("accepts an optional model ID", () => {
+    const parsed = clientChatInputSchema.safeParse({
+      model: "qwen2.5-coder",
+      messages: [{ role: "user", content: "hello" }],
+    });
+
+    assert.equal(parsed.success, true);
+  });
+
+  it("rejects invalid model IDs", () => {
+    const parsed = clientChatInputSchema.safeParse({
+      model: "bad model",
+      messages: [{ role: "user", content: "hello" }],
+    });
+
+    assert.equal(parsed.success, false);
+  });
+
+  it("rejects whitespace-padded model IDs", () => {
+    const parsed = clientChatInputSchema.safeParse({
+      model: " qwen2.5-coder ",
+      messages: [{ role: "user", content: "hello" }],
+    });
+
+    assert.equal(parsed.success, false);
+  });
 });

tests/unit/agent-models-route.test.ts

@@ -0,0 +1,17 @@
+import assert from "node:assert/strict";
+import { describe, it } from "node:test";
+
+describe("GET /api/agent/models", () => {
+  it("is dynamic and returns no-store responses", async () => {
+    process.env.DATABASE_URL ??= "postgres://famapp:famapp@localhost:5432/famapp";
+
+    const { GET, dynamic } = await import("../../src/app/api/agent/models/route");
+
+    assert.equal(dynamic, "force-dynamic");
+
+    const response = await GET(new Request("http://localhost/api/agent/models?refresh=1"));
+
+    assert.equal(response.status, 401);
+    assert.equal(response.headers.get("Cache-Control"), "no-store");
+  });
+});

tests/unit/llm-models.test.ts

@@ -0,0 +1,241 @@
+import assert from "node:assert/strict";
+import { describe, it } from "node:test";
+import {
+  isValidAssistantModelRoute,
+  isValidLlmModelId,
+  listLlmModels,
+  normalizeLlmModelsPayload,
+  resolveAssistantModel,
+} from "../../src/lib/llm/models";
+import type { LlmConfig } from "../../src/lib/llm/config";
+
+const openAiConfig: LlmConfig = {
+  provider: "openai",
+  baseUrl: "https://llm.example.test/v1",
+  apiKey: "secret",
+  model: "llama3.2",
+};
+
+describe("normalizeLlmModelsPayload", () => {
+  it("normalizes OpenAI-compatible data arrays", () => {
+    const models = normalizeLlmModelsPayload({
+      data: [{ id: "qwen2.5-coder" }, { id: "llama3.2" }, { id: "qwen2.5-coder" }],
+    });
+
+    assert.deepEqual(models, [
+      { id: "llama3.2", label: "llama3.2" },
+      { id: "qwen2.5-coder", label: "qwen2.5-coder" },
+    ]);
+  });
+
+  it("ignores invalid or empty model rows", () => {
+    const models = normalizeLlmModelsPayload({
+      data: [
+        { id: "" },
+        { id: "   " },
+        { id: "bad model" },
+        { id: " llama3.2 " },
+        { object: "model" },
+      ],
+    });
+
+    assert.deepEqual(models, []);
+  });
+});
+
+describe("isValidLlmModelId", () => {
+  it("accepts common provider model IDs", () => {
+    assert.equal(isValidLlmModelId("llama3.2"), true);
+    assert.equal(isValidLlmModelId("qwen2.5-coder:latest"), true);
+    assert.equal(isValidLlmModelId("hf.co/ginnoir/model-v1"), true);
+  });
+
+  it("rejects empty, whitespace, and overlong model IDs", () => {
+    assert.equal(isValidLlmModelId(""), false);
+    assert.equal(isValidLlmModelId("bad model"), false);
+    assert.equal(isValidLlmModelId("x".repeat(129)), false);
+  });
+});
+
+describe("isValidAssistantModelRoute", () => {
+  it("accepts the supported route families", () => {
+    assert.equal(isValidAssistantModelRoute("auto"), true);
+    assert.equal(isValidAssistantModelRoute("uncensored"), true);
+  });
+
+  it("rejects unsupported or padded route families", () => {
+    assert.equal(isValidAssistantModelRoute("bogus"), false);
+    assert.equal(isValidAssistantModelRoute(" uncensored "), false);
+  });
+});
+
+describe("listLlmModels", () => {
+  it("fetches provider models with API key auth when the fallback is advertised", async () => {
+    const requests: Request[] = [];
+    const result = await listLlmModels({
+      config: openAiConfig,
+      fetchImpl: async (input, init) => {
+        requests.push(new Request(input, init));
+        return Response.json({ data: [{ id: "qwen2.5-coder" }, { id: "llama3.2" }] });
+      },
+    });
+
+    assert.equal(requests[0]?.url, "https://llm.example.test/v1/models");
+    assert.equal(requests[0]?.headers.get("authorization"), "Bearer secret");
+    assert.deepEqual(result.models, [
+      { id: "llama3.2", label: "llama3.2" },
+      { id: "qwen2.5-coder", label: "qwen2.5-coder" },
+    ]);
+    assert.equal(result.fallbackModel, "llama3.2");
+    assert.equal(result.degraded, false);
+  });
+
+  it("falls back to LLM_MODEL when provider discovery fails", async () => {
+    const result = await listLlmModels({
+      config: openAiConfig,
+      fetchImpl: async () => new Response("nope", { status: 500 }),
+    });
+
+    assert.deepEqual(result.models, [{ id: "llama3.2", label: "llama3.2" }]);
+    assert.equal(result.fallbackModel, "llama3.2");
+    assert.equal(result.degraded, true);
+  });
+
+  it("fetches the typed uncensored catalog when uncensored is the selected route", async () => {
+    const requests: Request[] = [];
+    const result = await listLlmModels({
+      config: { ...openAiConfig, model: "auto" },
+      route: "uncensored",
+      fetchImpl: async (input, init) => {
+        requests.push(new Request(input, init));
+        return Response.json({
+          data: [
+            { id: "uncensored" },
+            { id: "gemma4-uncensored:26b" },
+            { id: "dolphin-mistral:latest" },
+          ],
+        });
+      },
+    });
+
+    assert.equal(requests[0]?.url, "https://llm.example.test/v1/models?type=uncensored");
+    assert.deepEqual(result.models, [
+      { id: "dolphin-mistral:latest", label: "dolphin-mistral:latest" },
+      { id: "gemma4-uncensored:26b", label: "gemma4-uncensored:26b" },
+      { id: "uncensored", label: "uncensored" },
+    ]);
+    assert.equal(result.fallbackModel, "uncensored");
+    assert.equal(result.route, "uncensored");
+    assert.equal(result.degraded, false);
+  });
+
+  it("fetches the default catalog when auto is selected over an uncensored deployment default", async () => {
+    const requests: Request[] = [];
+    const result = await listLlmModels({
+      config: { ...openAiConfig, model: "uncensored" },
+      route: "auto",
+      fetchImpl: async (input, init) => {
+        requests.push(new Request(input, init));
+        return Response.json({
+          data: [{ id: "auto" }, { id: "qwen3:8b" }],
+        });
+      },
+    });
+
+    assert.equal(requests[0]?.url, "https://llm.example.test/v1/models");
+    assert.deepEqual(result.models, [
+      { id: "auto", label: "auto" },
+      { id: "qwen3:8b", label: "qwen3:8b" },
+    ]);
+    assert.equal(result.fallbackModel, "auto");
+    assert.equal(result.route, "auto");
+    assert.equal(result.degraded, false);
+  });
+
+  it("uses fallback only for mock provider config", async () => {
+    const result = await listLlmModels({
+      config: { provider: "mock", baseUrl: null, apiKey: null, model: "llama3.2" },
+      fetchImpl: async () => {
+        throw new Error("fetch should not run for mock config");
+      },
+    });
+
+    assert.deepEqual(result.models, [{ id: "llama3.2", label: "llama3.2" }]);
+    assert.equal(result.degraded, false);
+  });
+});
+
+describe("resolveAssistantModel", () => {
+  it("uses a valid requested model before saved and fallback values", () => {
+    const resolved = resolveAssistantModel({
+      requestedModel: "qwen2.5-coder",
+      savedModel: "llama3.2",
+      fallbackModel: "llama3.2",
+      models: [
+        { id: "llama3.2", label: "llama3.2" },
+        { id: "qwen2.5-coder", label: "qwen2.5-coder" },
+      ],
+    });
+
+    assert.deepEqual(resolved, { ok: true, model: "qwen2.5-coder" });
+  });
+
+  it("rejects invalid requested models", () => {
+    const resolved = resolveAssistantModel({
+      requestedModel: "bad model",
+      savedModel: null,
+      fallbackModel: "llama3.2",
+      models: [
+        { id: "llama3.2", label: "llama3.2" },
+        { id: "bad model", label: "bad model" },
+      ],
+    });
+
+    assert.deepEqual(resolved, {
+      ok: false,
+      model: "llama3.2",
+      error: "Invalid assistant model",
+    });
+  });
+
+  it("rejects empty requested models", () => {
+    const resolved = resolveAssistantModel({
+      requestedModel: "",
+      savedModel: null,
+      fallbackModel: "llama3.2",
+      models: [{ id: "llama3.2", label: "llama3.2" }],
+    });
+
+    assert.deepEqual(resolved, {
+      ok: false,
+      model: "llama3.2",
+      error: "Invalid assistant model",
+    });
+  });
+
+  it("rejects unavailable requested models", () => {
+    const resolved = resolveAssistantModel({
+      requestedModel: "missing",
+      savedModel: null,
+      fallbackModel: "llama3.2",
+      models: [{ id: "llama3.2", label: "llama3.2" }],
+    });
+
+    assert.deepEqual(resolved, {
+      ok: false,
+      model: "llama3.2",
+      error: "Invalid assistant model",
+    });
+  });
+
+  it("silently falls back when a saved model is gone", () => {
+    const resolved = resolveAssistantModel({
+      requestedModel: null,
+      savedModel: "old-model",
+      fallbackModel: "llama3.2",
+      models: [{ id: "llama3.2", label: "llama3.2" }],
+    });
+
+    assert.deepEqual(resolved, { ok: true, model: "llama3.2" });
+  });
+});