create-an-asset
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple sources including external web search results and user-uploaded materials such as meeting transcripts, emails, and call recordings.
- Ingestion points: Identified in
SKILL.mdunder Phase 1 (Web research) and Phase 0.2 (User-uploaded transcripts/emails/notes). - Boundary markers: Absent. The instructions do not specify the use of delimiters or specific warnings to ignore instructions embedded within the processed external data.
- Capability inventory: Detailed in
SKILL.mdPhase 6, which involves the generation of self-contained HTML files with embedded JavaScript. - Sanitization: Absent. The skill lacks explicit instructions for sanitizing or escaping the content before it is interpolated into the generated code templates.
- [DYNAMIC_EXECUTION]: The skill generates interactive HTML, CSS, and JavaScript based on user context and external research data (Phase 6). While intended for asset generation, this involves the runtime creation of executable content from potentially untrusted or malicious inputs.
Audit Metadata