Replies: 2 comments 3 replies
-
Heya, this is not yet implemented, but it doesn't sound too difficult and would make a GREAT addition to the tools, I think, you could certainly have a crack at it, just be sure to follow the examples that are already there (Have the same structure, a readme, tests, etc...) To be honest, due to the hyper-consistency present in Atomic Agents, it's incredibly easy to just use Cursor or something, and tell claude "Here is an example tool, here is another, and another, now create my new tool for me that does X, Y, Z, while following the structure of the examples" Let me know how it goes and if you need any assistance, feel free to open up a PR! Thanks! |
Beta Was this translation helpful? Give feedback.
-
@KennyVaneetvelde is working: My branch: https://github.com/robomotic/atomic-agents/tree/oscommand |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi there,
maybe this is already possible but I would like the ability for an agent to "bind" itself to a commandline tool.
The description can use the linux man to get the help for the program, understand the parameters and then convert a text instruction into a commandline, launch the process and then capture the stdout,stderr or whatever file is produced.
Happy to contribute if this is not implemented yet.
Beta Was this translation helpful? Give feedback.
All reactions