Cc Safety Net

CC Safety Net is a safety layer for AI coding agents that stops destructive shell commands before they execute. It hooks into the agent's command path, inspects each proposed bash command, and returns a clear denial with a reason when the command would destroy local state or expose secrets. Typical blocks include a hard git reset, discarding working tree changes, cleaning untracked files, force pushing without lease protection, and recursive deletion of the home or project directory. It is for developers who let Claude Code, Codex, Cursor or another supported CLI run commands in their workspace and do not want to rely on written guidelines alone. The tool analyses command intent rather than matching brittle text patterns, and it catches shell wrapper forms such as commands passed through bash. Version 2 added 132 secret protection rules that block reads and writes of environment files, SSH keys, cloud credential stores and coding agent tokens before command analysis begins, active at every safety level. Teams can also write custom rules, and a bundled skill can inspect a project, draft a rulebook, verify it and fix errors. The site publishes no pricing and no paid plans; the product is presented as a free install, with documentation covering the full list of blocked operations and how to tune protected paths. When a command is blocked, the denial message tells the agent to ask the user for explicit permission and have the user run it manually, which keeps a human in the loop for genuinely necessary destructive operations.