The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local file system, specifically file names, directory structures, and file metadata. A malicious actor could attempt to influence the agent's decision-making or instruction adherence by creating files with names that contain hidden instructions.
Ingestion points: File and directory metadata gathered via ls, find, and du commands in user-specified target directories (SKILL.md).
Boundary markers: The skill uses a structured 'Organization Plan' and mandates a confirmation step ("Ready to proceed?") before executing changes.
Capability inventory: The skill possesses the ability to create directories (mkdir), move files (mv), and delete files (Step 5, Step 6 in SKILL.md).
Sanitization: No explicit sanitization or escaping of shell metacharacters in filenames is defined, relying instead on manual user review of the proposed plan.
[COMMAND_EXECUTION]: The skill uses several shell utilities including ls, find, du, sed, uniq, md5, mkdir, and mv. These commands are necessary for the skill's stated purpose of analyzing and reorganizing file systems. Instructions emphasize logging moves and preserving modification dates as safe practices.

file-organizer