The short answer
Yes, you can deploy and run an app on Antideploy without opening a dashboard. Paste one sentence into your coding agent and click Approve on one link. After that, your agent deploys, sets variables, reads logs, checks health and creates a database. You tell it what you want in plain words.
Say this to your agent, once
Set this project up to deploy on Antideploy. Fetch https://antideploy.com/agent.md and follow it.
What you still do in a browser
Some actions stay in the Antideploy console on purpose. They are irreversible, they spend your money, or they widen access. Your agent can tell you to do them, but it cannot do them itself.
-
The first approval. You open one link and click Approve. That connects your agent to your account.
-
Deleting an app or a database. An agent can create a database, but dropping one or deleting an app is done by you, in the console.
-
Attaching your own domain. You add it on the app's Domains tab, then add one DNS record at your registrar.
-
Adding money to the AI wallet. Your agent can create an AI key and give you the top-up link, but only you can add money.
-
Revoking access. You can see and revoke your agent's token at any time at antideploy.com/tokens.
What to say to your agent
Once it is connected, you do not need to know the API. Say what you want. These are examples.
-
"Deploy the latest change." It sends the new version and tells you when it is live.
-
"Is my app up?" It reads the app's health and tells you whether it is running, and when it last checked.
-
"Show me the last 100 log lines." It reads what the app printed, newest last. You can ask for up to 300 lines.
-
"The last deploy failed. Fix it." It reads the failure reason and the logs, changes your code and deploys again.
-
"Add a Postgres database." It creates one and writes the connection details into your
.envfile without printing them. -
"Set STRIPE_SECRET_KEY to the value in my .env." It saves the value into the encrypted store. Values are write-only, so it can list names but never read a value back.
-
"Run /api/cron/nightly every day at 3am UTC." It creates a scheduled job that calls that path.
-
"Scan my live app for security problems." It starts a full scan and reads the result.
Why this works
Your agent reads one machine-readable page that lists every endpoint, request shape, error code and limit. It makes every call itself, so there is nothing for you to click. Errors say what is wrong, whether to retry and what to do instead, so the agent can act on them.
When you do want to look, the console is there. It shows the same facts: deploy history, logs, metrics, health, databases and settings. It is optional.