floor-plan
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [Prompt Injection] (SAFE): The use of instructional markers like 'IMPORTANT' and 'NEVER' are natural directions for output formatting (ensuring the correct code fence is used) rather than attempts to bypass AI safety guardrails.
- [Metadata Poisoning] (SAFE): The author field contains promotional text for a browser extension and a link to Gitbook documentation. This is informational and does not contain deceptive instructions intended to alter agent behavior.
- [Remote Code Execution] (SAFE): There are no scripts, installation commands, or remote execution patterns present in the file.
- [Data Exposure] (SAFE): The skill does not request access to sensitive file paths, credentials, or attempt to exfiltrate data.
Audit Metadata