FICHA · MANJARO

ai-jail

Sandbox for AI coding agents (bubblewrap on Linux, sandbox-exec on macOS)

  • cli-tool
  • CLI
  • SECURITY
  • Dev
  • Launchable
  • Runs in terminal
official+codex · reviewed · May 30, 2026 description in en

Description

Automated coding tools can be run inside a sandbox that limits file system and process access. This is useful for reducing risk when AI agents inspect or modify code on Linux or macOS.

Sandboxing reduces exposure but does not make untrusted code safe by itself. Review allowed paths, network assumptions, and generated changes before trusting the result.

How to run

ai-jail

Commands: ai-jail

Permissions

Permissions not analysed for this source yet.