brand-setup
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to fetch an external website URL and its associated stylesheets for style extraction. This establishes a surface for indirect prompt injection, where a malicious website could embed instructions within CSS content to manipulate the agent.
- Ingestion points: Step 2 requires the agent to
Fetch [URL]andDownload each one [stylesheet] into styles/. - Boundary markers: The instructions do not provide delimiters or specific warnings to the agent to ignore instructions that might be embedded within the external CSS files.
- Capability inventory: The agent is granted capabilities to perform network fetches, file system writes (writing
[brand]-design-system.md,index.html, andlogo.svg), and local file listing. - Sanitization: There is no evidence of sanitization or filtering to ensure that the content extracted from the stylesheets consists only of design tokens rather than executable instructions.
Audit Metadata