Muse buys, your store sells
Shopping and payments were among Muse’s launch connectors (Shopify, Walmart, Sephora, Best Buy, Stripe, PayPal). A connector puts your catalogue and checkout in that flow.
Muse connectors
Muse, Meta’s personal AI agent, does tasks across people’s apps through connectors. Build your product’s tools once in Flow and have them ready for Muse, ChatGPT and Claude.
Muse’s connector program is in review-based early access. Facts here are from Meta’s announcements and press coverage.
Book dinner for 4 Friday, somewhere we haven’t been
opentable · find_tables
Lupa has a table at 8:15 and you haven’t been. Book it?
What you can build
Shopping and payments were among Muse’s launch connectors (Shopify, Walmart, Sephora, Best Buy, Stripe, PayPal). A connector puts your catalogue and checkout in that flow.
Search, hold and book from one request.
Reservations and delivery on request.
Notes, docs and tasks updated for the user.
Brand examples are illustrations of what could be built, not those companies' products.
The usual way vs. Flow
Building it yourself~6 weeks
Muse has published no SDK yet; a firsthand report says its form takes a live API or a hosted MCP server (unconfirmed).
With Flow Frameworkan afternoon
import { tool } from "@flow-engineer/flow"; import { z } from "zod"; export default tool({ name: "free_slots", description: "List open tables for a party size and time.", input: z.object({ when: z.string(), guests: z.number() }), run: async (q) => api.tables.available(q), });
How Flow works
Get started
Run npx @flow-engineer/flow new my-agent, or ask Claude Code or Codex to set it up. You get one folder: the agent’s instructions, its tools, its skills and its tests.
Each job is a TypeScript function with a typed input that calls your API: check an order, book a slot, find a deal. No protocol code.
Run npx @flow-engineer/flow test. Simulated users try the scenarios that matter, including people who change their mind or ask for something your agent should refuse.
Flow serves your tools as an MCP server on HTTPS with each user’s sign-in, which is what assistants like ChatGPT and Claude connect to today.
Apply through Meta’s Muse connector program with your live server. Meanwhile the same project is live in ChatGPT and Claude.
$ npx @flow-engineer/flow new my-agent $ cd my-agent $ npx @flow-engineer/flow test ✓ 24 scenarios passing
Guide
Muse is Meta’s personal AI agent, launched on 8 September 2026 and running on Meta’s Muse Spark model. It connects to a person’s apps, such as email and calendar, and carries out tasks across their accounts.
Connectors are how Muse reaches other products. Meta describes two kinds: directory connectors that developers submit and Meta reviews, and custom connectors that Muse sets up for a single user. Meta said over 1,500 developers applied in the program’s first week.
Apply through Meta’s connector program; Meta reviews each submission for function, security and legal requirements and tests it end to end. Meta has not published an SDK, fees or review times. Having a working, signed-in API or MCP server ready is the practical first step, and it also gets you into ChatGPT and Claude.
Not officially confirmed. Meta’s Connect recap describes connecting via an API or an MCP server, and one firsthand report says the form accepts a hosted MCP server. We will update this page when Meta publishes docs.
Launch partners reported by TechCrunch include Stripe, Shopify, Shop Pay, PayPal, Best Buy, Gap, Sephora, Walmart, Wayfair, Expedia, GitHub, Granola and Notion, with Instacart coming.
Meta has said Muse will be free for a large number of tokens and plans to earn a small fee on transactions.
Meta’s personal AI agent, launched 8 September 2026. It connects to a person’s apps and does tasks across them.
Apply through Meta’s Muse connector program, which reviews and tests each submission. Have a live API or MCP server with user sign-in ready; Flow builds that from your tools.
Meta has not published one as of October 2026.
Not officially confirmed. Meta’s materials mention connecting via an API or an MCP server.
Directory connectors are submitted by developers and reviewed by Meta. Custom connectors are set up by Muse for one user, without review.
Yes. A Flow project serves the same tools as an MCP server for ChatGPT and Claude today.
One open-source TypeScript project. The same agent on WhatsApp, iMessage, Telegram, ChatGPT and more.