The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes bash_tool to perform directory scanning and file management tasks such as ls, find, mkdir, and mv. These commands are restricted to the target directory specified by the user or the standard upload path (/mnt/user-data/uploads/).
[SAFE]: The skill implements a rigorous 'Confirmation Gate' (Phase 4), ensuring that no destructive or permanent changes (renaming, moving, deleting) occur without explicit user approval of the proposed reorganization plan.
[PROMPT_INJECTION]: The skill processes untrusted file content (Phase 2) to determine categorization. While this presents a surface for indirect prompt injection, the skill is designed to summarize and propose actions rather than execute code based on file contents, and standard LLM guardrails for reading data apply.
[DATA_EXPOSURE]: Access is limited to local file paths and uploaded user data, which is necessary for the skill's primary function of file auditing and organization.

folder-organizer