AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
Why coding agents should run in a sandbox, what sandy isolates and how, animated walkthroughs of the boundary at work, and how to get running.
Developer toolsShell · PythonGitHub @rappdwOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
Isolation for your coding agents... Run your coding agent in a Docker sandbox with filesystem isolation, network isolation, and per-project credential sandboxes.
Taken from the repository README
Loading the full README. The saved excerpt is shown below.
sandy — an isolated sibling for your coding agents
New to sandy? sandy, explained is a short guided walkthrough: what it isolates, animated examples of the boundary at work, and how to get running.
2.0 renames the container user and home from claude to sandy. /home/claude is baked into virtualenv shebangs, GOPATH, PYTHONUSERBASE and npm/cargo metadata, so every sandbox created by 1.x must be migrated — sandy refuses to launch against one rather than limping into it and failing later in ways that look like broken packages.Read the full README ↗The full README is unavailable right now. Read it at the original link. The saved excerpt is shown below.
sandy — an isolated sibling for your coding agents
New to sandy? sandy, explained is a short guided walkthrough: what it isolates, animated examples of the boundary at work, and how to get running.
2.0 renames the container user and home from claude to sandy. /home/claude is baked into virtualenv shebangs, GOPATH, PYTHONUSERBASE and npm/cargo metadata, so every sandbox created by 1.x must be migrated — sandy refuses to launch against one rather than limping into it and failing later in ways that look like broken packages.Read the full README ↗