The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/ledger.py utilizes subprocess.run to invoke its own CLI during internal self-testing. This execution is restricted to the script's own path using the current Python interpreter, presenting no risk of external command injection.
[INDIRECT_PROMPT_INJECTION]: The evolution cycle ingests external data from local logs and git history to generate proposals. While this is an attack surface for indirect injection, the skill enforces a mandatory human-in-the-loop review process where all AI-generated changes are proposed as pull requests and must be manually merged by the repository owner.

skill-evolution