designprompt
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION] (LOW): The style files in the
styles/directory contain<role>blocks that instruct the agent to adopt specialized personas (e.g., 'expert frontend engineer'). While used for the skill's primary purpose, these blocks are a form of instruction override. - [EXTERNAL_DOWNLOADS] (LOW): The skill documentation and scripts facilitate the collection of design specifications from
designprompts.devand suggest installing standard UI libraries likelucide-react. - [Indirect Prompt Injection] (LOW): The skill ingests data from an external website without sanitization or strict boundaries. 1. Ingestion points: Scraped files in the
styles/directory. 2. Boundary markers: Style files lack delimiters or instructions to ignore embedded commands. 3. Capability inventory: The agent is expected to generate and refactor frontend code. 4. Sanitization: No validation of scraped content is performed.
Audit Metadata