file-organizer
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill uses standard Unix commands (
ls,find,du,mv) to perform file management tasks as described in the metadata. The instructions include safety protocols such as providing a clear plan to the user and requiring confirmation before executing changes or deleting files.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local file system, which constitutes a potential injection surface.\n - Ingestion points: Reads directory contents and file metadata through
lsandfindcommands (SKILL.md).\n - Boundary markers: The instructions do not define technical delimiters to isolate file metadata from the agent's prompt context.\n
- Capability inventory: The skill has capabilities to create directories, move files, and propose file deletions across user-specified directories (
SKILL.md).\n - Sanitization: There is no mention of sanitizing or validating filenames or content before they are analyzed by the agent.\n
- Verdict justification: The risk is considered safe because all destructive actions require explicit user confirmation and the data is limited to the user's own local environment.
Audit Metadata