CAGE (Command Agent Guard Environment)
Visit WebsiteGitHub RepoAI SafetyIdea / Pre-seed (very early open-source project; newly created, no traction signals like stars)Unknown (not mentioned in provided repo metadata)
Description
Default-deny command protection for AI coding agents that blocks dangerous shell commands at execution time. Designed to work with tools like Claude Code, Codex, Cursor, and Aider to mitigate prompt-injection and unsafe command execution.
Founders
le12484 (GitHub owner; individual maintainer, real name not provided)
Discovered
January 26, 2026
Added to Database
January 26, 2026
Notes
As AI coding agents gain autonomy, shell-level guardrails become a critical control point; CAGE’s default-deny approach can reduce catastrophic command execution risk regardless of the agent framework. Positioned as a lightweight security layer that can plug into multiple popular agentic coding workflows.