The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes filesystem auditing and maintenance commands such as df, du, git, trash, cp, and ln. The instructions mandate specific safety protocols, such as verifying physical free space on the correct volume (/System/Volumes/Data) and preferring tool-native cleanup commands over manual directory deletion.
[INDIRECT_PROMPT_INJECTION]: The workflow involves the agent dynamically discovering and executing cleanup commands based on "current official documentation" and tool help outputs.
Ingestion points: Tool help menus, external documentation retrieved at runtime, and metadata from Git worktrees.
Capability inventory: Shell command execution and filesystem modification.
Boundary markers: Absent; the skill does not explicitly instruct the agent to ignore instructions embedded in the documentation it reads.
Sanitization: No explicit sanitization or validation of the retrieved command strings is performed before execution, relying on the agent's internal reasoning.
[DATA_EXPOSURE]: The skill performs targeted filesystem scans to identify large files and caches. Access is scoped to metadata (file sizes and paths) necessary for the primary cleanup purpose, and no network exfiltration patterns or hardcoded credentials were detected.

system-cleanup