Do you can recommand any existing MCP server to mange linux VM

I'm having a ubuntu VM or oracle OCI, and whenever i have to solve some problem inside the server i keep jugling back and forth from chatgpt to the server to keep executing the code which lead to wasting time and not giving right informations to chatgpt sometimes and i get overwhealmed, so i really would love to see if ther's already an existing solution so i can make things simple on my self, also my work really not that difficult, all my applications runs on docker/coolify

AIWeb said:

just install shell-gpt on the server. it lets you query openai directly from the command line so you dont have to keep switching windows. you can even have it execute the commands it generates if you use the shell flag. really saves a lot of time when messing with docker configs. another option is using vs code with remote ssh and an extension like continue.dev, keeps everything in one window.

Click to expand...

Informative, i will try to look over this

tbh if you are using those free tier oci instances be careful with running too many background tools since the ram is pretty limited on the AMD micro shapes. shell-gpt is nice but sometimes just setting up a split terminal in vscode or using something lightweight like aichat works better without slowing the server down. also coolify has a built-in terminal in the dashboard for basic docker stuff which might save you some ssh hassle anyway.

Colonel Potter said:

tbh if you are using those free tier oci instances be careful with running too many background tools since the ram is pretty limited on the AMD micro shapes. shell-gpt is nice but sometimes just setting up a split terminal in vscode or using something lightweight like aichat works better without slowing the server down. also coolify has a built-in terminal in the dashboard for basic docker stuff which might save you some ssh hassle anyway.

Click to expand...

they give you 4oCUPs and 24GB RAM tho i'm just going to have my ai agent on my own computer while running the SSH to the server so it can control it, but the heavy lifiting on the resources would be not even on my machine since i'm using an API connection, it's not a local model, i'm not sure if we're talking in the same direction

if you want the agent on your local machine but running commands on the server, maybe look at open interpreter... you can configure it to use ssh pretty easily. since you have that 24gb arm instance resource limits arent really going to be an issue anyway