# EmbodiChain task assets Downloadable task bundles used by `embodichain.data.assets.task_assets`. Files preserve their source bytes; sizes and SHA-256 checksums are in [manifest.json](manifest.json). | Archive | Purpose | MD5 | | --- | --- | --- | | [PourCupAssets.zip](PourCupAssets.zip) | Five runtime meshes for cup pouring (cup, mug wall, receiver, table, floor). | `f97857b639189be4daf717a873ab84cf` | | [PourCupDemoData.zip](PourCupDemoData.zip) | Verified MuJoCo trajectory, SPH reference snapshot, videos, preview images and validation reports. | `8eb96cb756ac08efe403417f50ad2bf2` | Python examples: ```python from embodichain.data import get_data_path cup = get_data_path("PourCupAssets/cup.glb") trajectory = get_data_path("PourCupDemoData/trajectory.pt") ``` The current fluid task generates its initial water fill in memory. `particles_30.npz` is historical validation evidence, not an initial-fill asset. The demo archive preserves provenance reports from their original recordings. Reports are reference evidence, not claims that a later source revision was re-simulated. Source: [EmbodiChain 0f132ef68](https://github.com/DexForce/EmbodiChain/tree/0f132ef6821915df1ff3ebed73a425b9e8776869).