skills/utarn/rus-zip/rus-zip-backup/Gen Agent Trust Hub

rus-zip-backup

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill inherently processes untrusted data from the filesystem and database engines, creating a potential surface for indirect prompt injection.
  • Ingestion points: The skill reads data from local directories (e.g., /srv/data) and database dumps generated by tools like pg_dump or mysqldump to create archives. It also ingests .zrus archives during restoration.
  • Boundary markers: The instructions strictly mandate verification steps, including SHA-256 checksum validation and integrity testing using rus-zip test before any restore operation.
  • Capability inventory: The skill utilizes file system operations (rm, rsync, cp), database interactions (pg_dump, psql), and archive manipulation tools (rus-zip).
  • Sanitization: The skill enforces a "Restore Invariant" that prohibits extracting archives directly onto live data, requiring a staging directory (e.g., /tmp/restore-staging) and manual confirmation before final placement.
  • [COMMAND_EXECUTION]: The skill defines sequences of shell commands to orchestrate backup and restoration tasks.
  • Evidence: Examples include the use of rus-zip, pg_dump, psql, sha256sum, rsync, and rm to handle data collection, compression, verification, and cleanup.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 02:26 AM