Universal chat layer for building bots and agents

A unified TypeScript SDK for building chat bots with type-safe handlers, JSX cards, and multi-platform support—powered by Vercel

$npm install chat
general
TypeScriptbot.ts
bot.onNewMention(async (thread) => {  await thread.subscribe();  await thread.post("Sure! Here's a quick summary...");});
bot.onReaction(async (thread, reaction) => {  await thread.post(`Thanks for the ${reaction.emoji}!`);});
bot.onSubscribedMessage(async (thread, msg) => {  await thread.post("Checking now...");});
2M
Weekly downloads
2.3K
GitHub stars
98+
Contributors
15+
Adapters

The Platform-Agnostic Chat Toolkit

The open-source chat toolkit designed to help developers build chat bots that run on Slack, Teams, Google Chat, Discord, WhatsApp, and more.

  • Multi-platform supportShip to every chat platform from one codebase, without per-platform rewrites.
  • Event-driven by designReact to mentions, reactions, and replies with handlers scoped to each thread.
  • Type-safe by defaultStrict types for adapters, handlers, and JSX cards, checked at build time.
bot.ts
import { Chat } from "chat";import { createSlackAdapter } from "@chat-adapter/slack";import { createRedisState } from "@chat-adapter/state-redis";
export const bot = new Chat({  userName: "mybot",  adapters: {    slack: createSlackAdapter(),  },  state: createRedisState(),});
bot.onNewMention(async (thread) => {  await thread.subscribe();  await thread.post("Hello! I'm listening to this thread now.");});
bot.onSubscribedMessage(async (thread, message) => {  await thread.post(`You said: ${message.text}`);});

Chat SDK Core

A unified API for building event-driven chat bots. Listen for mentions, subscribe to threads, and post rich cards across multiple platforms.

Visit Documentation

Supports

+ more adapters

Scale with confidence

Plug Chat SDK into an entire ecosystem designed for AI-native chat experiences that scale.

AI SDKBuild AI agents with streaming, tool calls, and structured outputs.
$ npm i ai
Vercel AI GatewayAccess 100+ models with one API key and no markup.
$ npm i ai
Vercel SandboxBuild knowledge agents with persistent filesystems to search, index, and read files.
$ npm i @vercel/sandbox
Workflows NEWBuild durable chat agents that suspend, resume, and survive function timeouts.
$ npm i workflow

Build with Chat SDK today

Visit Documentation
$npx create-chat-sdk@latest

Vercel Connect Guide

Authenticate a Slack bot with Vercel Connect instead of stored secrets.

import { createSlackAdapter } from "@chat-adapter/slack";import { connectSlackAdapter } from "@vercel/connect/chat";
const slack = createSlackAdapter({  ...connectSlackAdapter("slack/acme-slack"),});

Knowledge Agent Template

Answer questions from synced docs and repos with file-system search.

const savoir = createSavoir({  apiUrl: process.env.SAVOIR_URL,  apiKey: process.env.SAVOIR_API_KEY,});
const { text } = await generateText({  model,  tools: savoir.tools,  maxSteps: 10,  prompt: "How do I configure auth?",});

Code Review Bot Guide

Review pull requests with sandboxed AI analysis on GitHub.

bot.onNewMention(async (thread, msg) => {  const { data: pr } = await octokit.pulls.get({    owner, repo, pull_number,  });  await thread.post("Starting code review...");  const review = await reviewPullRequest({    owner, repo,    prBranch: pr.head.ref,  });  await thread.post(review);});