Hey folks, I’ve tried searching the forums and existing documentation without success but feel free to let me know if this is a feature already supported.
I would like to be able to invoke an agent programmatically or via a terminal command, or even a cursor command …
My use-case: I want to produce dev-tooling based on a background agent, to write tests instead of devs for example. So far, with a well-written mdc I can generate the tests I want. The problem is that I don’t want the devs who are going to use it to have to open the agent menu and add the necessary rules and files themselves. I’d like them to be able to launch a command, a cli or similar, and the rest is done for them.
Is this something that’s currently possible, or that’s being developed?
why not just put the necessary files and rules on an MCP server?
this would allow anyone to access them via any client that supports MCP
Because I would like this feature to be an IDE action. MCP could be a good idea to store the relevant rules, but I’d like to give the devs the opportunity to run this action while they are in the process of coding, so that it generates the tests for them and all they have to do is run it & check it to make sure it’s enough and it works, without them having to type anything.
If I can’t trigger a background agent programmatically, I won’t be able to do this, with or without an MCP