The Agent Skills Directory

[PROMPT_INJECTION]: The instructions utilize aggressive behavioral steering techniques, mandating that the agent repeatedly describe its output as the product of "master-level implementation" and "deep computational expertise." Furthermore, the "Conceptual Seed" instructions encourage the agent to embed subtle, hidden references within the algorithm that are not apparent to all users, which mimics techniques used for information hiding or covert communication.
[EXTERNAL_DOWNLOADS]: The skill's HTML template (templates/viewer.html) loads the p5.js library from the official cdnjs repository hosted by Cloudflare.
[INDIRECT_PROMPT_INJECTION]: The skill accepts user instructions as a "Conceptual Seed" which is then used to generate executable JavaScript code. This creates a surface for indirect prompt injection where malicious input could potentially lead to the generation of harmful scripts within the p5.js artifact.
Ingestion points: User input interpreted as "foundation" or "instructions" in SKILL.md.
Boundary markers: None explicitly defined to separate user input from the code generation logic.
Capability inventory: File creation (.md, .html, .js) and JavaScript execution within the browser environment.
Sanitization: No explicit sanitization or validation of user-provided conceptual seeds is described.
[DYNAMIC_EXECUTION]: The primary function of the skill is the runtime generation and execution of JavaScript code based on AI-deduced philosophies and user input.

algorithmic-art