CC Safety Net 🤖 Agent Open source

What it does

CC Safety Net is a pre-execution guard for AI coding agents. Before a tool call runs, it inspects the command and blocks destructive operations and access to secrets. It parses what a command actually does, so wrapping it in bash -c or python -c, reordering flags, or otherwise disguising it does not slip past the check. It denies the call rather than sandboxing the process, and a broken config file never silently disables protection.

What is inside

Built-in rules block destructive Git and filesystem commands (git reset --hard, git push --force, rm -rf on dangerous targets, find -delete, PowerShell Remove-Item) and reads or writes to SSH keys, .env files, ~/.aws, and stored CLI credentials. Three presets (Standard, Strict, Paranoid) trade coverage against false positives, and a local web GUI lets you toggle individual rules, add allow or deny paths, and review what was blocked. Official rulebooks for Terraform, AWS, gcloud, and Azure add further blocks; a rulebook can only add restrictions, never remove built-in ones. Diagnostics include status, doctor, explain, and an on-machine audit log. A library API (checkCommand) lets you embed the analyzer in your own tools.

Works with

Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot CLI, OpenCode, Amp, and Google Antigravity, plus several other CLIs, on Windows, macOS, and Linux (some integrations are best-effort on Windows).

How to install

npx -y cc-safety-net@latest install

Maintenance and safety

MIT licensed and actively maintained, with CI and coverage badges. Requires Node.js 18 or higher. The audit trail records command decisions locally but not command output or prompts. It denies calls only; it does not set filesystem permissions, watch network egress, or contain a process, and its path matching is mostly POSIX.

Who should use it

Anyone running autonomous or semi-autonomous coding agents who wants a safety net against accidental data loss and secret exposure.