The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses common shell utilities such as ls, find, du, and md5 to analyze file structures. These operations are restricted to local file management and are controlled by instructions that require user approval before any changes are applied.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes file names and metadata from directories like 'Downloads', which can contain untrusted data. This inherent surface risk is mitigated by the skill's specific purpose and the instructions for the agent to log all actions and seek confirmation before proceeding with modifications.
Ingestion points: ls, find, and file commands in SKILL.md are used to gather data from the file system.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded content in file names.
Capability inventory: The agent uses mv, mkdir, and is instructed to ask for permission to delete files.
Sanitization: No explicit sanitization of file names is mentioned before processing.

file-organizer