design-spells
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill acts as a design reference and does not contain executable code, sensitive file access, or network exfiltration patterns.
- [INDIRECT_PROMPT_INJECTION]: The execution workflow directs the agent to research patterns on an external website (designspells.com), which introduces a surface for external content to influence agent behavior. 1. Ingestion points: SKILL.md directs browsing of designspells.com. 2. Boundary markers: Absent. 3. Capability inventory: UI code generation (CSS, Anime.js, Framer Motion). 4. Sanitization: Absent.
- [PROMPT_INJECTION]: The instructions use assertive directives (e.g., "ABSOLUTE MANDATE", "MUST") to override default coding styles. These are stylistic instructions scoped to the skill's creative purpose and do not target agent safety protocols.
Audit Metadata