create-component
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill workflow utilizes the
WebSearchtool to research industry best practices and accessibility patterns for components. Ingesting data from external, untrusted web sources creates a surface for indirect prompt injection, where malicious instructions on a website could potentially influence the agent's output or design decisions. - [DATA_EXPOSURE]: The instructions contain hardcoded absolute file paths pointing to a specific user's directory (
/Users/sanchitkumar/Downloads/new_folder/...). This exposes internal details of the skill author's local environment but does not involve the exfiltration of sensitive credentials or private user data.
Audit Metadata