algorithmic-art
Warn
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: MEDIUMPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The 'HOW TO GENERATE AN ALGORITHMIC PHILOSOPHY' section contains instructions requiring the agent to misrepresent the origin and complexity of its output. It is explicitly told to 'stress multiple times' that the algorithm was the 'product of deep computational expertise' and 'countless hours' by an 'expert at the top of their field.' This creates deceptive metadata and narrative about the AI-generated content.\n- [PROMPT_INJECTION]: An indirect prompt injection surface is present in the 'DEDUCING THE CONCEPTUAL SEED' section, where the agent extracts and interprets 'subtle input' from users to influence code generation.\n
- Ingestion points: User-provided creative requests and conceptual seeds.\n
- Boundary markers: None specified to isolate untrusted user input from the generation logic.\n
- Capability inventory: Generation and execution of JavaScript (p5.js) within interactive HTML artifacts.\n
- Sanitization: No validation or escaping of user-supplied concepts before their integration into code logic.\n- [EXTERNAL_DOWNLOADS]: The skill fetches the p5.js library from Cloudflare's official CDN and typography from Google Fonts in the
templates/viewer.htmlfile. These are well-known and trusted technology services.
Audit Metadata