The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill provides numerous Bash commands for Git management. These are standard operations for version control and do not include dangerous flags or attempts to execute arbitrary system commands outside of the Git context.
[INDIRECT_PROMPT_INJECTION] (LOW): As a Git skill, it facilitates the ingestion of external data from remote repositories (e.g., git pull, git clone). While this is an inherent attack surface for AI agents, the skill itself does not provide malicious instructions or encourage pulling from untrusted sources.
[REMOTE_CODE_EXECUTION] (SAFE): While the documentation mentions Git Hooks and tools like npm run lint, it does not automate the download or execution of remote scripts from untrusted URLs (e.g., curl | bash).

git-advanced