Any recommendations how to setup an agent that is sandboxed well enough to contribute?

We’re building Stashbase for this (repo: https://github.com/stashbase) You can run your coding agent in Docker-based sandbox with explicit workspace mounts and outbound network rules. It supports your own Dockerfile or image too, so you can keep whatever research tools and dependencies you need (default Dockerfile installs Claude Code and Codex). The access rules live in a TOML profile.

If you're on Linux you can use lxc-box to quickly spin up containers and VMs. I made a wrapper that creates them, installs codex and shares directories with the host in a single command, using a docker-like syntax