Quickstart: get manufacturing quotes from your AI agent

This quickstart connects your coding agent to the Maglev MCP server and gets real quotes for a CAD part. You need a Maglev account (free to create) and Claude Code or Codex.

Connect your agent

Claude Code (one command)
Claude Code (with a key)
Codex

Run this in a terminal. It opens a browser to sign in or sign up, saves a key to ~/.maglev/token, and adds Maglev to Claude Code:

curl -fsSL https://trymaglev.com/connect.sh -o connect.sh && sh connect.sh

If it prints MAGLEV WAITING, finish signing in and run the same command again. It resumes the same sign-in.

Create a key on the dashboard (API keys, then Create secret key), then run:

claude mcp add --scope user --transport http maglev https://trymaglev.com/mcp \
  --header "Authorization: Bearer YOUR_MAGLEV_KEY"

Create a key on the dashboard, then add Maglev to ~/.codex/config.toml:

[mcp_servers.maglev]
url = "https://trymaglev.com/mcp"
http_headers = { Authorization = "Bearer YOUR_MAGLEV_KEY" }

If you added Maglev before, delete the old [mcp_servers.maglev] block first.

Open your agent in a folder with CAD files

Maglev accepts STEP files (.step, .stp) and wire-harness specs (.json), up to 20 MB each.

cd my-robot-parts
claude

Ask for quotes

Tell the agent what you want, for example:

Get quotes for 10 of each part in this folder.

The agent calls make_start to upload the files, check_manufacturability to run each shop’s design check, and get_quotes to ask every shop that makes each part’s process. A four-part request usually answers in 10 to 20 seconds. The agent shows you every offer.

Order, only if you approve

If you want to order, the agent calls place_order first without a total. That returns the final total and charges nothing. The order goes through only after you approve that exact amount.

Your requests, offers and orders also appear on the dashboard, with each part’s offers side by side.

Next steps

MCP tools

What each of the 16 tools does and the arguments it takes.

Use the REST API

The same flow over HTTP, with request and response examples.

⌘I

Get started

MCP server

How-to

Open data

API reference

Quickstart: get manufacturing quotes from your AI agent

Next steps

MCP tools

Use the REST API

​Next steps

MCP tools

Use the REST API

Next steps