fixing-metadata
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists exclusively of markdown instructions and guidelines. No executable scripts, shell commands, or automated tools are included in the provided files.- [INDIRECT_PROMPT_INJECTION]: The skill provides guidance for processing external data within HTML metadata. It includes specific instructions to 'escape and sanitize any user-generated or dynamic strings,' which helps prevent injection attacks in the target application.- [DATA_EXFILTRATION]: The instructions highlight the risk of external debuggers exposing sensitive URLs, explicitly advising the agent not to submit private or staging content without authorization.
Audit Metadata