nano-banana-prompting
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's instructions and reference materials are entirely focused on guiding users through the creative process of prompt engineering for image generation. No hidden commands or malicious intent were detected.
- [PROMPT_INJECTION]: The skill was evaluated for potential prompt injection or jailbreak attempts. The instructions follow a standard conversational workflow and do not include patterns designed to bypass safety filters or override agent behavior.
- [DATA_EXFILTRATION]: Analysis confirmed that the skill does not access sensitive local file paths (such as credentials, SSH keys, or environment variables) and contains no network-related commands for external data transmission.
- [REMOTE_CODE_EXECUTION]: The skill does not include any scripts, package installations, or remote code download patterns. It operates purely through structured text prompts and local reference files.
Audit Metadata