Agent sandbox
An isolated environment where an AI agent runs code and commands without access to your network, secrets or real files.
Coding agents and data-analysis agents execute what the model writes. In a sandbox (a container, a micro-VM or a hosted code interpreter) a wrong or injected command can only damage a throwaway copy. Good sandboxes also block outbound network traffic by default, which removes the way out from the lethal trifecta, and pass in only the files and credentials the task needs.
Related terms
- Lethal trifecta: An AI agent that has private data, reads untrusted content and can send data out can be tricked into leaking that data.
- Coding agent: An agent that reads a repository, edits files, runs commands and tests, and opens pull requests.
- Excessive agency: An AI agent with more tools, permissions or autonomy than its task needs, so a mistake or an injected instruction can do real damage.
