First, an introduction:
monkeycode is an online development platform by Changting Baizhiyun, a subsidiary of Alibaba Cloud.
What’s different from ordinary coding tools:
With its built-in model, you can chat, control servers, and more.
The platform includes a container server within each chat session, with the following specs:
2 CPU cores, 8GB RAM
The platform also uses a points system:
Each day you get 30 minutes of free model conversations
using the basic free model.
You can also use advanced models by spending points.
You can earn points by checking in daily or completing tasks.
As far as I can tell, running the server itself doesn’t cost any points,
and conversations within the daily 30-minute allowance don’t cost points either.
However, calling tools through the model does use points.
If you don’t chat for a certain period (three days), the session gets cooled down or paused.
Apparently, this pause doesn’t stop projects already running inside the server,
but I haven’t tested this precisely.
I tested by installing an astbot project:
If you go back to the chat page, you can restore access even without chatting.
ps: I’m not sure if “no chat” means the domain access is paused (the page shows “cooled down”), or the server itself is shut down.
Server info:
It comes with a built-in SSH terminal, but port 22 is blocked—you can only use the web-based terminal.
AI chat can help install projects, but it seems everything is fed through review context.
The AI will refuse to install server management panels like bt panel, 1p, etc.
The page will warn that this behavior is risky and has been reported to the cloud, and repeated actions may deduct points.
You can install GitHub projects normally.
Large models will assess projects for security risks.
Risky projects might not deploy properly.
Most server control actions need to go through the AI model; the built-in SSH terminal doesn’t seem able to execute scripts, etc.
Server basic info:
I’ll leave the rest for everyone to explore yourself.
How to use
With affiliate link:
Without affiliate link:
If you can, feel free to use my affiliate—if not, that’s totally fine 
To start a new chat session,
just tell the model to check the server.
It’ll automatically bring up the server deployment.
The system recommends choosing the first one.
Just wait for it to start
and then you can ask the model to do server-related tasks.
If anyone finds fun or creative ways to tinker, let me know! 