feat: general cooking-question chatbot on the recipes homepage

The existing recipe chat (RecipeChatPanel) is tightly scoped to one
recipe — recipeId is required and its system prompt embeds that recipe's
full ingredient/step context. Added a sibling, not a modification: a new
floating assistant on the recipes list page for questions that aren't
about any specific recipe (technique, substitutions, timing, food safety).

New /api/v1/ai/cooking-chat route (same quota/rate-limit/model-resolution
pattern as recipe-chat, minus the recipe lookup) and CookingAssistantPanel
component (same floating-button + Sheet UI). Positioned at bottom-24
rather than bottom-6 so it doesn't sit in the same band as the recipe
list's floating bulk-action bar during select mode.

Verified locally: asked a real cooking question through the actual UI,
got a correctly-formatted markdown answer with no recipe context leaking
in (confirmed the request payload only carries the question, no recipeId).

apps/web/app/(app)/recipes/page.tsx

@@ -7,6 +7,7 @@ import { eq, desc, asc, and, ilike, or, inArray } from "@epicure/db";
 import { RecipesHeader } from "@/components/recipe/recipes-header";
 import { RecipesEmptyState } from "@/components/recipe/recipes-empty-state";
 import { RecipesGrid } from "@/components/recipe/recipes-grid";
+import { CookingAssistantPanel } from "@/components/recipe/cooking-assistant-panel";
 import { getMessages } from "@/lib/i18n/server";
 
 export const metadata: Metadata = {};
@@ -139,6 +140,8 @@ export default async function RecipesPage({ searchParams }: { searchParams: Sear
           )}
         </div>
       )}
+
+      <CookingAssistantPanel />
     </div>
   );
 }

apps/web/app/api/v1/ai/cooking-chat/route.ts

@@ -0,0 +1,50 @@
+import { NextRequest, NextResponse } from "next/server";
+import { z } from "zod";
+import { generateText } from "ai";
+import { requireSession } from "@/lib/api-auth";
+import { applyRateLimit } from "@/lib/rate-limit";
+import { withAiQuota, resolveAiConfigOrError } from "@/lib/ai/ai-error";
+import { getModelConfigForUseCase } from "@/lib/ai/resolve-user-key";
+import { resolveModel } from "@/lib/ai/factory";
+import { getUserPrivateBio, buildUserBioContext } from "@/lib/ai/user-bio";
+
+const Schema = z.object({
+  question: z.string().min(1).max(500),
+});
+
+const LANG: Record<string, string> = { en: "English", fr: "French" };
+
+export async function POST(req: NextRequest) {
+  const { session, response } = await requireSession();
+  if (response) return response;
+
+  const body = await req.json() as unknown;
+  const parsed = Schema.safeParse(body);
+  if (!parsed.success) {
+    return NextResponse.json({ error: "Validation error" }, { status: 400 });
+  }
+
+  const limited = await applyRateLimit(`rl:ai:${session!.user.id}`, 30, 60);
+  if (limited) return limited;
+
+  const [configResult, privateBio] = await Promise.all([
+    resolveAiConfigOrError(() => getModelConfigForUseCase(session!.user.id, "text")),
+    getUserPrivateBio(session!.user.id),
+  ]);
+  if (!configResult.ok) return configResult.response;
+  const model = resolveModel(configResult.data);
+  const bioContext = buildUserBioContext(privateBio);
+  const locale = (session!.user as { locale?: string }).locale ?? "en";
+  const lang = LANG[locale] ?? "English";
+
+  const result = await withAiQuota(session!.user.id, session!.user.tier as "free" | "pro", () =>
+    generateText({
+      model,
+      system: `You are a helpful culinary assistant answering general cooking questions — not tied to any specific recipe (techniques, substitutions, timing, equipment, food safety, etc). If a question has nothing to do with cooking or food, politely redirect. Keep answers under 200 words. Respond in ${lang}.${bioContext}`,
+      prompt: parsed.data.question,
+    })
+  );
+  if (!result.ok) return result.response;
+
+  return NextResponse.json({ answer: result.data.text });
+}