The Agent Skills Directory

[SAFE]: The skill's instructions are focused on providing accurate technical documentation for Three.js r160+. All code examples demonstrate standard use cases such as GLTF loading and character state transitions using crossfading.
[PROMPT_INJECTION]: Evaluated instructions for bypass or override markers. Directives like 'ALWAYS' and 'NEVER' are used strictly in a technical context to prevent common implementation errors (e.g., forgetting to update the animation mixer) and do not attempt to manipulate the agent's safety constraints.
[DATA_EXFILTRATION]: No sensitive file access or unauthorized network operations were identified. Code snippets use relative paths for local assets (e.g., 'character.glb'), and external links point exclusively to well-known, official documentation sites (threejs.org).
[REMOTE_CODE_EXECUTION]: The skill uses standard library imports and does not involve the execution of remote scripts or unverified dependencies. The imports target the official Three.js package.
[DYNAMIC_EXECUTION]: No usage of eval, exec, or runtime code generation from untrusted sources was found. Dynamic logic is limited to standard Three.js animation API calls.
[OBFUSCATION]: The content was analyzed for hidden characters, Base64 strings, and homoglyphs. No obfuscation techniques were detected.

threejs-impl-animation