The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes common Unix utilities such as ls, find, du, md5, mkdir, and mv to perform its core functions of file analysis and organization. These operations are performed on local directories with intended user oversight.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted filenames and file contents to categorize and organize data. This presents a potential attack surface where malicious filenames could attempt to influence the agent's behavior or exploit shell command parsing.
Ingestion points: Filenames and file metadata (type, size, hash) collected from the user's filesystem via commands like ls, find, and file.
Boundary markers: No explicit delimiting markers are defined for the file content, though the skill instructions emphasize requiring user confirmation before executing destructive or organizational actions.
Capability inventory: The skill possesses filesystem read/write capabilities, including directory creation (mkdir), file movement (mv), and file deletion (in the context of duplicate removal).
Sanitization: The skill relies on the agent's underlying shell execution environment to correctly handle and escape special characters or control sequences within filenames.

file-organizer