Vouch request: shayanjm

[shayanjm]

What do you want to work on?

I want to address #776 . I'm the Chief Data & AI Officer of Thoughtworks, an Nvidia Advanced Technology Partner. We're quite interested in openshell-based sandboxing for claw-like agents in the enterprise. We (and most of our clients) support multiple model providers, and we want the ability to have sandboxes use different inference providers while keeping a single stable in-sandbox endpoint like inference.local.

Note: We would love to keep collaborating to make OpenShell (and stuff in the OpenShell ecosystem) as enterprise-ready as possible. This is probably the first of many such contributions we'd like to make!

Why this change?

This would let one gateway manage multiple sandboxes without exposing provider credentials to agents, requiring direct provider egress from each sandbox, or introducing extra concepts for users. My goal is to contribute a minimal, upstream-friendly implementation that preserves the existing OpenShell architecture while enabling multi-provider deployments.

Checklist

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