input-types
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is entirely instructional, focusing on best practices for HTML form design. It does not contain executable code, sensitive data access, or network calls to untrusted locations. All external references target reputable informational resources related to frontend standards.
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes user-provided source code and HTML templates, which constitutes a data ingestion surface. While malicious instructions could be embedded in this external content to attempt to influence the agent, the risk is minimal as the skill's capabilities are limited to providing analysis and structural code fixes, without access to sensitive credentials, system files, or external network communication.
- Ingestion points: Review of templates, server-rendered HTML, and shared components (SKILL.md).
- Boundary markers: The skill does not define specific delimiters to separate user code from agent instructions.
- Capability inventory: Limited to code auditing, explanation, and text-based markup modification.
- Sanitization: No specific sanitization or filtering of the ingested HTML content is mentioned.
Audit Metadata