Build with your own agent
Connect Claude Code or another coding agent to your Appentic account, and let it create projects, add services, deploy and troubleshoot.
Appentic has an interface built for coding agents at mcp.appentic.com, served over MCP (Model Context Protocol, the standard way coding agents connect to tools). Connect an agent to your account and, from your own editor or terminal, it can do much of what you would otherwise do on the dashboard: create a project, price a machine before adding it, add a database, push code, deploy, read the logs and run commands on the machine.
What you need
- An Appentic account with credits. See Connect Claude Code.
- A coding agent that speaks MCP over HTTP. Claude Code is the one we test against.
How it differs from Athena
Athena is an AI engineer inside Appentic. You describe an app in plain words on the project's Athena page and it builds it, using credits from your balance.
Connecting your own agent is the other way round. Your agent runs wherever you run it, on your own model subscription, and Appentic is the platform it builds on. You write the prompts and see the code as it's written.
| Athena | Your own agent | |
|---|---|---|
| Where it runs | Inside Appentic | Your computer |
| Who pays for the model | Your Appentic credits | You, wherever your agent bills |
| Where you work | The project's Athena page | Your editor or terminal |
| What it builds | Feather apps | Anything Appentic can run |
Both end in the same place: an ordinary Appentic project, with the same services, logs and metrics. The code is yours either way.
What an agent can do
Once connected, an agent can:
- see your workspaces, your projects and what each one runs
- price a machine and a set of services before anything is added
- create a project, add its first machine, and add a web service, database, key value store, worker, cron job, storage, custom domain, firewall or volume
- deploy from your own GitHub repository or, if you don't have one, push to a private repository Appentic creates for the project
- deploy, follow the deploy, and read the build output and the app's logs
- check whether the app works, restart it, and read the deployed code
- push a fix to your own repository, once you allow it
- set environment variables
- run a command on the project's machine, such as a migration or a test run against the real services
Appentic refuses anything that costs money unless the agent confirms it, and never returns the value of an environment variable. See What an agent can do for every tool and its limits.
What it is not
It is not a public API. The interface exists for coding agents working on your behalf, and its tools change as the product does. If you want to build something against Appentic programmatically, write to us first.