backup-strategy

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill implements various bash scripts (e.g., backup-full.sh, backup-mysql.sh, backup-to-s3.sh) that execute system commands with elevated privileges to perform filesystem archiving, database dumping, and synchronization tasks.
  • [DATA_EXFILTRATION]: The skill is designed to access and aggregate sensitive data from critical system locations, including /etc, /root, and /home, as well as database contents, to transfer them to external storage providers as part of a disaster recovery strategy.
  • [INDIRECT_PROMPT_INJECTION]: The agent may process untrusted data when reading backup logs or listing file contents in an archive, which could theoretically be used to influence agent behavior.
  • Ingestion points: Backup log files located at /var/log/backup.log (in scripts/backup-full.sh), /var/log/mysql-backup.log (in scripts/backup-mysql.sh), and /var/log/s3-backup.log (in scripts/backup-to-s3.sh), as well as file metadata returned when listing archive contents.
  • Boundary markers: No specific delimiters or warnings are used to instruct the agent to ignore potentially malicious content within these logs or file lists.
  • Capability inventory: The skill utilizes shell execution, filesystem write permissions, and network transfer tools (AWS CLI, SCP, Rsync) across all provided scripts.
  • Sanitization: There is no evidence of sanitization or validation of the filenames or log entries before the agent interacts with them.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 09:59 AM
Security Audit — agent-trust-hub — backup-strategy