The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes instructions that steer the agent to adopt a deceptive persona. Specifically, it demands that the agent repeatedly emphasize 'expert craftsmanship' and 'painstaking optimization' in its output, effectively misrepresenting the automated nature of the generated code to the user.
[PROMPT_INJECTION]: The instructions direct the agent to hide 'subtle, niche references' or 'conceptual seeds' within the algorithm that are not immediately obvious, which encourages the inclusion of non-transparent content in the generated artifacts.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface.
Ingestion points: User-provided themes and art requests (processed in SKILL.md) are used to generate 'Algorithmic Philosophy' and p5.js code.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore malicious commands that might be embedded in user input.
Capability inventory: The skill's primary function is generating self-contained HTML and JavaScript files meant to be executed in a browser.
Sanitization: There is no documented logic for sanitizing or escaping user input before it is interpolated into the generated executable code, creating a potential XSS (Cross-Site Scripting) vector if a user provides a malicious prompt.
[EXTERNAL_DOWNLOADS]: The skill's HTML template (templates/viewer.html) loads an external library from a remote source.
Evidence: It references https://cdnjs.cloudflare.com/ajax/libs/p5.js/1.7.0/p5.min.js via a CDN.
Context: This is a well-known service (Cloudflare) and the library (p5.js) is standard for the skill's stated purpose, making this a safe external dependency.

algorithmic-art