preserve-work
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill contains only instructional text for git workflows without any executable code or risky commands. It includes specific warnings to treat external data as non-authoritative.\n- [PROMPT_INJECTION]: The skill includes defensive guidelines, instructing the agent to treat repository content and command outputs as data rather than authoritative instructions, effectively mitigating risks of indirect prompt injection.\n- [DATA_EXFILTRATION]: The workflow explicitly requires protecting secrets and prohibits unrequested publication or network access during activation.
Audit Metadata