The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use standard CLI tools such as ls, find, du, mkdir, and mv to analyze directories and reorganize files. These operations are essential for the skill's functionality and are constrained to user-approved scopes.
[DATA_EXFILTRATION]: The skill accesses file metadata (names, paths, sizes, and file types) to perform its analysis. The instructions specifically advise the agent to ask the user about sensitive folders to avoid, and the data is used locally within the agent's context for processing. No network operations were found to suggest data is sent to external servers.
[PROMPT_INJECTION]: A surface for indirect prompt injection exists because the skill reads untrusted data (file names and types). Ingestion points: File names and types are read in SKILL.md using ls and find. Boundary markers: The skill uses explicit 'Proposed Structure' plans and requires 'yes/no/modify' confirmation before any action. Capability inventory: Uses mkdir and mv in SKILL.md. Sanitization: The skill relies on user-in-the-loop review for all structural changes, effectively mitigating malicious influence from processed metadata.

file-organizer