The Agent Skills Directory

[PROMPT_INJECTION] (LOW): The skill generates a file named AGENTS.md containing instructions for other agents. These instructions use aggressive overriding language such as 'CRITICAL, MANDATORY, NO EXCEPTIONS' and 'THIS IS NOT OPTIONAL'. While used here for a benign git workflow, this style of language is a primary pattern for prompt injection intended to bypass or override an AI's core safety or operational constraints.
[COMMAND_EXECUTION] (SAFE): The skill executes git init, which is the standard command for its stated purpose of initializing a repository. No signs of arbitrary or malicious command execution were found.

here-be-git