housekeep-files

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a localized utility for organizing document trees via copy operations. It does not perform network operations, request elevated privileges, or implement persistence mechanisms.
  • [DATA_EXPOSURE]: The included scripts (scripts/housekeep-files.ps1 and scripts/housekeep-files.sh) contain a security filter that identifies sensitive file types, such as .env, .pem, .p12, and various key formats, using a regular expression. The logic ensures these files are only indexed in place rather than copied, mitigating the risk of credential proliferation.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes file system metadata (filenames and directory structures), which is an inherent surface for indirect prompt injection. The implementation successfully mitigates this risk through defensive programming.
  • Ingestion points: Local filenames and directory paths are read during the inventory phase.
  • Boundary markers: The scripts utilize literal paths for all file operations and unit separators (ASCII 31) during plan parsing to prevent command injection via malicious filenames.
  • Capability inventory: The skill is restricted to file system read, copy, and SHA-256 hashing operations.
  • Sanitization: All catalog entries are JSON-escaped before being written to the manifest file, ensuring that special characters in filenames are handled safely.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 11:27 PM