algorithmic-art

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill employs 'expert' role-play instructions and repetitive behavioral steering to influence the model's output style, specifically directing the model to prioritize its own creative 'philosophy' over user-provided constraints.
  • [EXTERNAL_DOWNLOADS]: The skill's HTML template fetches the p5.js library from the trusted Cloudflare CDN (cdnjs.cloudflare.com).
  • [PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection. 1. Ingestion points: User-provided artistic requests are ingested to generate a philosophy and corresponding code. 2. Boundary markers: Absent; the instructions explicitly state that user input 'should not constrain creative freedom'. 3. Capability inventory: The skill generates HTML and JavaScript artifacts that are executed in the user's local browser context. 4. Sanitization: No mechanism is defined to sanitize user-provided descriptions before they are used to determine the logic and parameters of the generated p5.js sketch.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 03:24 AM