OpenShell
Policy-enforced sandbox runtime for running autonomous AI agents
What it is
OpenShell runs AI agents inside isolated sandboxes where kernel controls limit file access, system calls, and network connections according to a declared policy. Credentials are injected only into requests headed for approved endpoints, and proposed policy changes are checked with formal verification so risky new access waits for human approval. It ships a CLI, a gateway control plane deployable on Kubernetes via Helm, and SDKs for Python, TypeScript, Go, and Rust. Platform and security teams use it to let coding agents and other autonomous tools install packages and call APIs without exposing secrets or the wider network.
