mac-software-storage-cleanup

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local bash scripts to perform system audits and cleanup tasks. • scripts/report_sizes.sh uses du and find to inventory application directories. • scripts/list_priority2_candidates.sh uses du to inspect application support folders. • scripts/cleanup_priority1.sh uses rm -rf to delete contents within ~/Library/Caches and ~/Library/Developer/CoreSimulator after user confirmation.
  • [SAFE]: The skill implements safety measures to prevent accidental data loss. • The SKILL.md instructions explicitly mandate waiting for user confirmation before executing any cleanup scripts. • Deletion is limited to specific, low-risk cache directories. • Bash scripts use set -euo pipefail for error handling and handle filenames with null terminators to prevent injection via specially crafted file names.
  • [PROMPT_INJECTION]: The skill processes filenames from the local filesystem which acts as an ingestion point for untrusted data (Indirect Prompt Injection). • Ingestion points: scripts/report_sizes.sh and scripts/list_priority2_candidates.sh read filenames from the system. • Boundary markers: Absent in prompt interpolation. • Capability inventory: File deletion via scripts/cleanup_priority1.sh. • Sanitization: Scripts use null-terminated strings for shell safety, though content is presented raw to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 10:41 AM