OpenShell
A safe, private runtime for fleets of autonomous AI agents that enforces file, process and network access through kernel-level sandboxing and policy checks.
OpenShell is a runtime for running fleets of autonomous AI agents that can read files, install packages, call APIs and use credentials without unrestricted access to a user's data, secrets or network. Each agent runs in an isolated sandbox, and a declared policy states what it may touch.
The kernel instruments file access, system calls and network connections so every action is checked against that policy at runtime, and credentials are added only to requests bound for approved endpoints, never exposed to the agent directly. Policy changes are checked with formal verification before being applied, flagging risky new access such as reaching a new host or API method for human review.
It installs via a shell script on Linux, Apple Silicon macOS, or Windows with WSL2, using Docker, Podman or host virtualization, and includes a gateway and supervisor for managing sandboxes, providers and policy, with a Helm chart for Kubernetes deployment.