Sandbox-to-Sandbox Communications enables controlled communication between OpenShell sandboxes. This supports multi-agent workflows where one sandbox runs an agent and another hosts a tool, service, model, database, or ephemeral resource.
This supports patterns such as:
- Sandbox A with long running agent running tool on Sandbox B
- Sandboxes being used to deploy ephemeral or long-running resources (such as local LLMs)
The goal is to make cross-sandbox communication explicit, policy-controlled, observable, and ergonomic, without requiring direct unmanaged reachability between sandbox environments.
Sandbox-to-Sandbox Communications enables controlled communication between OpenShell sandboxes. This supports multi-agent workflows where one sandbox runs an agent and another hosts a tool, service, model, database, or ephemeral resource.
This supports patterns such as:
The goal is to make cross-sandbox communication explicit, policy-controlled, observable, and ergonomic, without requiring direct unmanaged reachability between sandbox environments.