[Feature request] Remote Servers SSH

[EasyArt]

It would be nice to add Remote Servers Credentials password and username or sshkey and then set a cron job for a specific server.

[fccview]

Oh wow that's such a cool idea!
Not sure how doable it is, I'll definitely investigate!

[tracure1337]

Yes, this needs an agent mode. Otherwise it's beautiful but just using normal cron is like 10k x more efficient ;-)

[fccview]

Yes, this needs an agent mode. Otherwise it's beautiful but just using normal cron is like 10k x more efficient ;-)

I am not going to add an agent mode to something that has full access to your entire system with no constraints sorry.

Also not sure about that tho, it catalogues logs neatly, allows you to run scripts with a click, with live logs, let's you switch users from the UI easily and you can just see what's scheduled for when at glance.

Yes you can do all that via command line obviously, but it's definitely not more efficient that way is it 😆

This uses native crontab btw ♥️

[freedbygrace]

Yes, this needs an agent mode. Otherwise it's beautiful but just using normal cron is like 10k x more efficient ;-)

I am not going to add an agent mode to something that has full access to your entire system with no constraints sorry.

Also not sure about that tho, it catalogues logs neatly, allows you to run scripts with a click, with live logs, let's you switch users from the UI easily and you can just see what's scheduled for when at glance.

Yes you can do all that via command line obviously, but it's definitely not more efficient that way is it 😆

This uses native crontab btw ♥️

I would second an agent mode because architecturally, the agent is making outbound connections to check for policy or work and executes according to the rights the agent service has been deployed with. The agent does not need to be an agent that opens a port or ports from the client perspective and then the server makes server initiated connections. That would be inefficient. But if the server can receive on 443 https tls for example, then agents are just checking for work and have to establish a handshake with the server before any jobs or instructions are exchanged.

Might you reconsider?

[fccview]

Yes, this needs an agent mode. Otherwise it's beautiful but just using normal cron is like 10k x more efficient ;-)

I am not going to add an agent mode to something that has full access to your entire system with no constraints sorry.
Also not sure about that tho, it catalogues logs neatly, allows you to run scripts with a click, with live logs, let's you switch users from the UI easily and you can just see what's scheduled for when at glance.
Yes you can do all that via command line obviously, but it's definitely not more efficient that way is it 😆
This uses native crontab btw ♥️

I would second an agent mode because architecturally, the agent is making outbound connections to check for policy or work and executes according to the rights the agent service has been deployed with. The agent does not need to be an agent that opens a port or ports from the client perspective and then the server makes server initiated connections. That would be inefficient. But if the server can receive on 443 https tls for example, then agents are just checking for work and have to establish a handshake with the server before any jobs or instructions are exchanged.

Might you reconsider?

I may have been a bit harsh with my first response..

My current concern with an agent mode is ensuring it can be deployed with the principle of least privilege and that it doesn't create new, unmonitored attack vectors, especially since the current implementation leverages existing, trusted native tools.

I do agree that an agent-based architecture it's still more secure than managing remote authentication storing SSH credentials centrally, that's for sure.. I'll look into how an agent could be implemented safely with constrained access/permissions. Not promising anything 😄