WaveletWorldModel

Generates a world model representation from state inputs using discrete wavelet transforms (DWT) to capture multi-resolution temporal and spatial features.

The skill appears coherent and limited to computing wavelet-based compressed state vectors. Before installing: (1) verify the skill's provenance (source is listed as unknown and there's no homepage), (2) run the included Python script in an isolated/sandbox environment (or a virtualenv) and install dependencies from trusted registries (pip install numpy PyWavelets) with pinned versions, (3) inspect the code yourself (it's short and readable) and confirm you will not feed it sensitive secrets or personally identifiable data, and (4) if you plan to integrate this into an automated agent, ensure inputs are validated and that dependency installation is controlled. If you need higher assurance, ask the publisher for a source repository or signed release artifacts.

Type: OpenClaw Skill Name: wavelet-worldmodel-skill Version: 1.0.0 The skill bundle is benign. All files (metadata, skill instructions, Python script, and README) align with the stated purpose of generating a world model using discrete wavelet transforms. The Python script uses standard scientific computing libraries (`numpy`, `pywt`), performs no file system access beyond dependency loading, no network calls, and no execution of external commands. There are no signs of data exfiltration, malicious execution, persistence, or prompt injection attempts in the markdown files.