The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of Bash scripts to perform disk audits and cleanups. It executes standard macOS utilities such as df, du, find, stat, and sort. It also invokes cleanup commands for popular developer tools like brew, uv, pip, npm, and docker.
[DATA_EXFILTRATION]: No network activity was detected. The skill reads file metadata (size, modification date, path) and presents it to the user via the AskUserQuestion tool for confirmation before any deletion. There is no evidence of data being sent to external servers.
[PROMPT_INJECTION]: The skill contains 'Self-Evolving' instructions, directing the agent to fix and update the SKILL.md file if it encounters issues. While this instructs the agent to modify its own logic, it is presented as a functional requirement for tool maintenance rather than an attempt to bypass safety or ethical guardrails.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of local filenames when scanning for large files or triaging the Downloads folder. Because the agent is instructed to update the SKILL.md file based on its execution experience, there is a theoretical surface where a maliciously named file could influence the agent's self-modification process. However, the scripts use proper shell quoting ("$f"), and the risk is assessed as low given the local context.

disk-hygiene