Vouch request: cosmicnet

[cosmicnet]

What do you want to work on?

Hello,

I'm a software engineer and mature student currently studying at Oxford. I've been setting up NemoClaw to take advantage of its security model. However, I've ended up patching it and would like to try to push the changes upstream.

It's common now to run OpenClaw with multiple models for inference. Such as local models for privacy, cheaper models for easier tasks, etc. The current OpenShell (NemoClaw) setup only allows for one model and one inference provide at any one time.

I saw that issue 203 was created to address this, however was closed without being implemented. The code looks as though many of the parts were put in place. My patch addresses issue 203, and also multiple models per inference provider, Ollama support, and OpenAI Codex support.

My changes can be viewed here:
https://github.com/cosmicnet/OpenShell/tree/203-multi-route-inference/lh

The changes are backward compatible.

I've tested this live in my NemoClaw sandbox and it seems to be working well.

I'm open to discussion on this. I was going to include full details in the pull request, but happy to share those now if you'd like.

Why this change?

This was a deal breaker for me being able to setup OpenClaw the way I needed to through NemoClaw. I'd assume others are in the same position.

Checklist

• I wrote this request myself (not AI-generated)
• I have read CONTRIBUTING.md

[johntmyers]

/vouch

[github-actions[bot]]

@cosmicnet has been vouched by @johntmyers.

You can now submit pull requests to OpenShell. Welcome aboard.

Please read CONTRIBUTING.md before submitting.