FluxyChatCookbook
Room history to AI messages
Convert FluxyChat room history into AI SDK-compatible messages with toAiMessagesFromRoom.
Room history → AI messages
Use FluxyChatClient.toAiMessagesFromRoom() to feed in-room chat history into agents, RAG pipelines, or Vercel AI SDK generateText / streamText calls.
Basic usage
import { FluxyChatClient } from "@fluxy-chat/sdk";
import { generateText } from "ai";
const client = new FluxyChatClient({
baseUrl: process.env.FLUXY_WORKER_URL!,
userId: "agent-bot",
token: memberJwt,
});
const messages = await client.toAiMessagesFromRoom("support-room-1", {
limit: 30,
includeSystem: false,
});
const result = await generateText({
model: openai("gpt-4o-mini"),
system: "Summarize the conversation for the next human agent.",
messages,
});
await client.createMessage("support-room-1", result.text);Options
| Option | Type | Description |
|---|---|---|
limit | number | Max messages to fetch (default 50) |
includeSystem | boolean | Include system-role messages |
mapUserDisplayNames | boolean | Prefix user messages with display names when available |
Agent invoke pattern
import { createTelemetryManager, createConsoleTelemetryIntegration } from "@fluxy-chat/sdk";
const telemetry = createTelemetryManager({ functionId: "room-summarizer" });
telemetry.register(createConsoleTelemetryIntegration());
const history = await client.toAiMessagesFromRoom(roomId);
// Pass history to your @fluxy-chat/agent runAgentLoop or external LLMRelated
- KB + RAG agent — retrieval over knowledge base
- AI agents handoff — human takeover from in-room AI
- Unified reasoning — cross-provider thinking controls