feat(agent): add voice input and photo attachments to assistant

Wire mic through Whisper-compatible transcriptions on LLM_BASE_URL.

Photos upload to MinIO and reach the vision model as base64 image_url parts.

src/modules/agent/messages.ts

@@ -0,0 +1,20 @@
+import { z } from "zod";
+
+export const clientChatAttachmentSchema = z.object({
+  type: z.literal("image"),
+  url: z.string().trim().min(1).max(2048),
+});
+
+export const clientChatMessageSchema = z.object({
+  role: z.enum(["user", "assistant"]),
+  content: z.string().trim().min(1).max(8000),
+  attachments: z.array(clientChatAttachmentSchema).max(3).optional(),
+});
+
+export const clientChatInputSchema = z.object({
+  stream: z.boolean().optional(),
+  messages: z.array(clientChatMessageSchema).min(1).max(40),
+});
+
+export type ClientChatAttachment = z.infer<typeof clientChatAttachmentSchema>;
+export type ClientChatMessage = z.infer<typeof clientChatMessageSchema>;