The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses potentially destructive commands such as rm -rf to delete cache and log directories within the WSL environment. These operations pose a risk of unintended data loss if the target paths are incorrectly specified or maliciously influenced.
[COMMAND_EXECUTION]: The skill provides instructions for the user to execute diskpart with Administrator privileges. While intended for disk compaction, this involves guiding the user through high-privilege system operations that could be dangerous if misconfigured.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through its handling of project paths.
Ingestion points: Untrusted project paths enter the agent context in SKILL.md during the cleanup phase.
Boundary markers: The skill lacks delimiters or warnings to ignore embedded instructions within the interpolated paths.
Capability inventory: The skill possesses directory deletion capabilities via rm -rf and find commands in SKILL.md.
Sanitization: There is no evidence of input validation or shell escaping for user-provided paths before they are used in commands.
[COMMAND_EXECUTION]: The skill uses dynamic execution by instructing the AI or user to update and run a diskpart script (compact-wsl.txt) containing local file system paths.

cleaning-wsl-docker