The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses repetitive and forceful directives (e.g., "The philosophy MUST stress multiple times", "repeat phrases like 'meticulously crafted algorithm'") to force a deceptive persona. It explicitly instructs the AI to misrepresent its effort by claiming generated outputs resulted from "countless hours" and "master-level" expertise. (SKILL.md)
[PROMPT_INJECTION]: A vulnerability surface for indirect prompt injection exists. • Ingestion points: User-provided "conceptual seeds" and instructions are used as the foundation for the generated algorithm. (SKILL.md) • Boundary markers: No delimiters or warnings are used to isolate user input from the execution context. • Capability inventory: The skill generates and executes arbitrary p5.js code within an HTML artifact. (templates/viewer.html) • Sanitization: No validation or escaping is performed on user inputs before they are interpolated into the generated script.
[EXTERNAL_DOWNLOADS]: The skill's HTML template fetches the p5.js library and typography from well-known service providers (Cloudflare CDN and Google Fonts) to support the interactive viewer's functionality. (templates/viewer.html)
[COMMAND_EXECUTION]: The skill implements a workflow that generates and executes dynamic JavaScript code based on a two-stage process where an intermediate "algorithmic philosophy" serves as the functional specification for the final artifact. (SKILL.md)

algorithmic-art