mac-software-storage-cleanup-cn
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local shell scripts to perform system auditing and maintenance tasks.
scripts/report_sizes.shandscripts/list_priority2_candidates.shuse standard utilities likedu,find, andawkto calculate directory sizes.scripts/cleanup_priority1.shusesrm -rfto clear contents within~/Library/Cachesand~/Library/Developer/CoreSimulatorafter auditing their size. - [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it ingests local file and directory names to generate storage reports, which are then processed by the agent.
- Ingestion points: Directory names and file paths retrieved from
/Applications,~/Applications, and~/Library/Application Supportviascripts/report_sizes.shandscripts/list_priority2_candidates.sh. - Boundary markers: None; filenames are listed as plain text in the output report without delimiters to prevent the agent from interpreting names as instructions.
- Capability inventory: The skill possesses the capability to delete files within specified cache directories using
rm -rfinscripts/cleanup_priority1.sh. - Sanitization: Filenames are processed by
awkanddubut are not sanitized or escaped before being presented to the agent in the report.
Audit Metadata