The Agent Skills Directory

[SAFE]: The skill is a technical guide for Three.js r185 WebGPU development and does not contain any malicious instructions or hidden payloads.
[INDIRECT_PROMPT_INJECTION]: The skill defines an architecture for processing external data (spawn packets, event intervals). The potential attack surface for indirect injection is well-mitigated:
Ingestion points: Data enters through structured inputs processed by scripts/trail-ring-oracle.mjs.
Boundary markers: The skill instructions emphasize deterministic seeds and immutable records to prevent data contamination.
Capability inventory: The skill and its referenced scripts have no access to network operations, file system writes, or shell command execution.
Sanitization: The provided script scripts/trail-ring-oracle.mjs implements rigorous type and range validation (e.g., verifying Number.isFinite, Number.isSafeInteger, and uint32 constraints) for all ingested numerical data, preventing exploitation through malformed inputs.
[REMOTE_CODE_EXECUTION]: The skill references standard, well-known libraries (three/webgpu, three/tsl) and provides its own local utility script. It does not attempt to download or execute code from untrusted external sources or perform silent package installations.
[DATA_EXFILTRATION]: There are no network-capable commands (e.g., curl, fetch) or instructions to access sensitive local files (e.g., .env, .ssh). All operations are focused on local GPU rendering and state management.

threejs-particles-trails-and-effects