Antideploy

BlogAny agent

How to deploy from any coding agent.

Two ways to deploy from any coding agent. Paste one sentence, or add the MCP server once. Both end at a live link.

How the Agent API works
  • One sentenceWorks in any agent that can read a web page
  • An MCP serverAdd it once with npx, nothing to install
  • One approvalOpen one link and click Approve
  • A live linkYour app on its own HTTPS address

The short answer

Any coding agent that can read a web page and make web requests can deploy to Antideploy. You have two ways to connect it. The first is one sentence that you paste into the agent. The second is the Antideploy MCP server, which you add once to an agent that supports MCP.

Start with the sentence. It covers more, because the agent follows a full procedure. Use the MCP server if you prefer tools that your agent can call by name.

Say this to your agent

Set this project up to deploy on Antideploy. Fetch https://antideploy.com/agent.md and follow it.

Way 1: paste the sentence

The sentence points your agent at a page called agent.md. That page is a procedure written for agents. The agent follows it from start to finish.

  1. Paste the sentence into your agent

    Open your project in the agent, paste the sentence and press Enter. The agent fetches agent.md and starts following it. Claude Code, Codex, Cursor, Antigravity and Windsurf all work this way.

  2. Approve one link

    The agent starts a device login and shows you a link and a short code. Open the link, check that the code matches, and click Approve. If you have no account yet, the link signs you in with Google or GitHub first and takes you straight back.

  3. Let the agent finish

    It creates the app, creates a Postgres database if your code needs one, deploys, and tells you where the app is live and on which account.

The agent needs to reach antideploy.com. If it runs in a sandbox that blocks the network, it stops and tells you which domain to allow. You can allow it, run the agent on your own machine, or upload the folder in the Antideploy console.

Way 2: add the MCP server

MCP is a standard way for an agent to call tools. Antideploy ships an MCP server as an npm package called antideploy-mcp. It runs with npx, so there is nothing to install by hand.

  1. Create the app and an API key

    In the Antideploy console, create the application and then create an API key for it. The key belongs to one application, so you never pass an application id.

  2. Add the server to your agent

    Add the server to your agent's MCP settings. For Claude Code it is one command:

    Claude Code
    claude mcp add antideploy \
    -e ANTIDEPLOY_API_KEY=ad_your_key \
    -- npx -y antideploy-mcp

    Other agents take the same three parts in their own settings: the command npx, the arguments -y antideploy-mcp, and the environment variable ANTIDEPLOY_API_KEY.

  3. Start a new session and ask it to deploy

    Your agent needs a new session to pick up the server. Then say "deploy this". The agent calls the deploy tool, gets back a task id, and checks deployment_status until the app is live.

The MCP server covers the essentials: deploying, checking status and setting variables. Everything beyond that is plain web requests, which every agent can already make. The server is open source under the MIT licence, on GitHub.

  • deploy. Sends your project and starts a deploy.
  • deployment_status. Follows a deploy until it succeeds or fails.
  • set_env and list_env. Sets variables and lists their names. Values are never returned.
  • api_info. Reads the full contract, every endpoint, request and limit.

Which one should I use?

Use the sentence if you want the agent to do everything: create the database, switch on storage or sign-in, set up cron jobs and read logs. You approve one link and nothing else.

Use the MCP server if your agent works best with named tools, or if you want one project key for one app that you can also use in a CI pipeline.

You can use both. An app deployed through MCP can be connected to a GitHub repository later without losing anything.

01Before you commit

Here's where it stops.

A platform that only tells you what it is good at is one you find the edges of in production. These are the ones to know before your first deploy.

  1. Apps sleep when idle

    On every plan, an app nobody is using goes to sleep and the next visit wakes it. The first request took 2.9 to 13.9 seconds in our measurements, depending on the stack. There are no always-on workers, so use a cron job or a webhook for background work.

  2. One region

    Everything runs in Singapore, with the database beside the app. Users far from Asia will see slower responses.

  3. One instance, no previews

    Each app is one instance with 1 shared vCPU and 1 GB of memory, and a Java app gets 1 dedicated vCPU and 2 GB. There are no preview deployments and no per-branch URLs.

  4. The disk forgets

    Anything written to local disk is gone on the next deploy. Apps that speak S3 get a bucket instead.

02Questions

Agent questions, answered.

Anything else? Write to us and a person answers.

support@antideploy.com
Which coding agents does it work with?

Claude Code, Codex, Cursor, Antigravity and Windsurf, and any agent that can read a web page and make requests. There is also an MCP server for any client that speaks MCP.

Do I need an Antideploy account before I start?

Not for the sentence. Paste it first. When the agent needs you, it shows a link, and if you have no account, signing in with Google or GitHub creates one. For the MCP server you create the app and key in the console, so you sign in first.

Is the token safe?

The token from the sentence is saved to ~/.antideploy/config.json on your machine and is never printed in the chat. You can revoke it at any time at antideploy.com/tokens. A project key is for one application and is built to survive being committed.

What if my agent cannot read web pages?

Then it cannot follow the sentence. Use the MCP server, or deploy from GitHub or by dropping a folder into the console.

Does the MCP server do everything the sentence does?

No. It covers deploying, status and variables. The sentence covers the whole platform, because the agent follows the full procedure and calls the same API directly.

+Keep reading

More guides, and the platform.

Each platform page covers one part of Antideploy: what it does, where it stops and the questions people ask. See the whole platform, or all the guides.

Facts on this page were checked against the live platform on 7 October 2026.

Deploy something. Start with one sentence.

Paste one sentence into your coding agent, click Approve once, and get a live link. No card, no trial clock.

Start from GitHub or a folder
Prompt copied Paste it into Claude Code, Codex or Cursor and press Enter.