The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill metadata contains an installation command that fetches the skill instructions from its official GitHub repository (raw.githubusercontent.com/majiayu000/claude-skill-registry). This is standard behavior for skill distribution and uses a well-known service.
[INDIRECT_PROMPT_INJECTION]: The skill provides templates for 3D charts and network graphs that process user-supplied data (labels, node IDs, values).
Ingestion points: The data prop in the BarChart3D component and the nodes/edges props in the NetworkGraph component.
Boundary markers: The sample code does not include specific boundary markers or 'ignore' instructions for the data being visualized.
Capability inventory: The skill focuses on rendering 3D graphics in a web context and does not include capabilities for file system modification, network exfiltration, or command execution within its visualization logic.
Sanitization: Standard React rendering is used, which provides inherent protection against basic script injection, though no custom natural language sanitization is present in the templates.
[SAFE]: The skill utilizes well-known, industry-standard libraries including Three.js, @react-three/fiber, and @react-three/drei. The code examples provided are functional, idiomatic, and contain no signs of obfuscation, credential harvesting, or unauthorized persistence.

3d-visualizer