CC Safety Net

CC Safety Net is an open-source, MIT-licensed pre-execution guard for AI coding agents, published by J Liew (kenryu42). It hooks into supported coding CLIs and blocks destructive commands and access to secrets such as SSH keys and .env files before the tool call runs. The latest release listed is v2.6.10.

What It Is

CC Safety Net is a safety layer for coding agents that run shell commands. It parses what a command actually does, so wrapping it in bash -c or python -c, or reordering flags, does not hide it. Safer variants such as git push --force-with-lease or git clean -n are allowed through. The docs state it is a pre-execution check, not a sandbox: it does not contain processes, set filesystem permissions, or watch network egress. A broken config file never blocks anything.

What It Blocks

Destructive commands include git reset --hard, git push --force, find -delete, and rm -rf on dangerous targets such as root or home. Secret protection covers SSH keys, .env files, ~/.aws, and coding-CLI credentials, both in the shell and in the agent's own read, edit, write, and search tools.

Policy and Team Setup

A local dashboard, started with npx cc-safety-net gui, lets users choose a Standard, Strict, or Paranoid preset and toggle rules. Official rulebooks add blocks for Terraform, AWS, gcloud, and Azure, and custom JSON rules are supported. Committing the .cc-safety-net/ directory shares policy across clones and cloud sessions. A checkCommand function can be called from Node.js without installing the hook.

Setup Path

It requires Node.js 18 or higher. One installer detects the coding agents on the machine and hooks into the ones the user picks, and a doctor command verifies protection. The site says decisions are made offline, audit logs stay in ~/.cc-safety-net, and there is no telemetry.