The Agent Skills Directory

[SAFE]: The skill implements a security-enhancing feature by providing a local shell script hook designed to block destructive git operations. The implementation is transparent and restricted to the user's local environment.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to install a shell script (scripts/block-dangerous-git.sh) and configure it as a PreToolUse hook within the Claude Code environment. This script runs locally to inspect and filter bash commands based on a hardcoded list of dangerous patterns. It uses standard utilities like jq and grep for command processing.
[DATA_EXFILTRATION]: No evidence of hardcoded credentials, sensitive file exfiltration, or unauthorized network communication was detected. The script correctly processes command input locally and does not transmit any data to external servers.
[SAFE]: The skill instructs the agent to modify .claude/settings.json or ~/.claude/settings.json to persist the guardrail hook. This modification is consistent with the skill's primary purpose and follows standard configuration practices for the target platform.

git-guardrails-claude-code