The Agent Skills Directory

[PROMPT_INJECTION]: No instructions were found that attempt to override agent behavior, bypass safety filters, or extract system prompts. The content is strictly technical and instructional.
[DATA_EXFILTRATION]: The code does not perform any network operations, access sensitive file paths, or contain hardcoded credentials. It operates entirely within the local browser context using standard WebGL and Canvas APIs.
[REMOTE_CODE_EXECUTION]: There are no patterns involving remote script downloads or execution. All code provided is self-contained and does not rely on external dependencies or package managers.
[OBFUSCATION]: No obfuscated content, such as Base64-encoded strings, zero-width characters, or homoglyphs, was detected. The code and GLSL shaders are presented in clear, readable formats.
[DYNAMIC_EXECUTION]: While the skill involves WebGL shader compilation at runtime, this is a standard and expected requirement for the stated purpose of fluid simulation. No unsafe general-purpose execution (e.g., eval or exec) is present.
[INDIRECT_PROMPT_INJECTION]: The skill processes numerical mouse and touch input for drawing purposes only. It does not ingest untrusted text data that could influence the agent's behavior.
[PRIVILEGE_ESCALATION]: The skill does not request or attempt to acquire elevated system privileges; it uses standard web platform APIs.

cursor-fluid-simulation