### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### The Feature

# What this feature is?

a subcommand like `lite launch <agent-tool>` , where `<agent-tool>` can be `claude`, `codex`, `opencode`, `openclaw`, `hermes` , it can use single litellm virtual key to launch the agent-tool with remote litellm enabled. It should also be able to obtain the list of all available models in the remote litellm instance. And let user to choose which one to use for the agent.

The behavior just simliar to `ollama launch`, but this is for using remote litellm.

# Why user need this?

Beside managing the remote litellm instance via `lite` command, user really has the following problem when using local agent with remote litellm instance:

1. It is difficult for user to configure different agent-tools for remote litellm. Each tools has their unique way to configure the remote litellm instance, even if there is documentation on litellm website, but it is still take user a lot of time to read and debug. `lite launch` would be a life-saver.

2. some of agent tools like `claude` may not be able to list all the models on litellm instan…