Git Push is Hard

A new approach to safely allow AI coding agents to run `git push` from sandboxed VMs uses a short-lived GitHub token injected by a proxy outside the sandbox, avoiding exposure of real credentials. The method, detailed in a technical guide, configures an nginx proxy on the host to attach a temporary GitHub App installation token to outgoing requests, enforcing rules like allowing push only to specific branches and denying force pushes. This pattern gives agents minimal, task-scoped permissions without granting full identity access.

Git Push is Hard If you run a coding agent inside a sandbox VM, the usual advice is simple: isolate the filesystem, restrict network access, don’t mount your home directory, don’t expose SSH keys or cloud credentials. But that leads to one question: without an SSH key, how would the agent run git push safely? You could give the VM an SSH key or a GitHub token. That basically diminishes the purpose of the VM, and gives the agent more access than it needs. So, don’t do this. AI agents had bad reputation https://openai.com/index/hugging-face-incident-and-the-road-ahead/ . Use a short-lived token instead Basically, configure a proxy outside the sandbox. Inside the VM, let git push talk to the remote with a scoped, synthetic credential. A custom proxy running outside the sandbox attaches the real GitHub OAuth token before forwarding the request — so the real token is invisible to the model. A GitHub App works well for this. The host keeps the app’s private key and creates a short-lived installation token for the repo. The VM uses that token to push its branch. The broker can enforce rules such as: - allow fetch - allow push agent/run-123 - deny push main - deny force push - deny other repos The agent does not need to hold the real GitHub credential at all. A practical guide In practice, the proxy can be as simple as an nginx docker process listening on 127.0.0.1:8082- 80/tcp . It injects gh auth token into outgoing requests as the auth header. Inside the VM, the push command looks like git push https://github.com/name/repo.git main . Getting an agent to use it doesn’t take much: just hint it. Something like “push via GitHub broker on host port 8082 instead of origin” works well — the agent figures out the rest on its own. Here’s a script that keeps that nginx conf fresh https://gist.github.com/soasme/0f93509db16e6a5a950cff26769ae11c : pull a short-lived token with gh auth token , write it into $XDG CONFIG HOME/ghe-proxy/nginx.conf as a Basic auth header, lock the file down to 0600 , then tear down and recreate the container — a plain restart would keep serving the old, stale token — and check docker inspect to confirm it actually came back up. The general pattern This is a useful way to think about agent permissions in general. A runtime does not need your identity. It needs a small set of capabilities for the current task. For git push , that capability is usually: this repo, this branch, for a short time.