Wildworld Dataset
Orient an agent on WildWorld’s action-conditioned ARPG frames, annotations, and WildBench when building or evaluating world models.
Overview
WildWorld Dataset is an agent skill for the Idea phase that documents the WildWorld action-conditioned ARPG dataset for generative world modeling research.
Install
npx skills add https://github.com/aradotso/trending-skills --skill wildworld-datasetWhat is this skill?
- 108M+ frames from a photorealistic AAA ARPG with 450+ semantically meaningful actions
- Per-frame skeletons, HP/stamina states, camera poses, and depth maps
- 29 monster species, 4 player characters, 4 weapon types, 5 distinct stages
- Clips up to 30+ minutes continuous for long-horizon world modeling
- WildBench benchmark evaluation path for generative ARPG world models
- 108M+ total frames
- 29 monster species and 5 distinct stages
Adoption & trust: 690 installs on skills.sh; 31 GitHub stars; 3/3 security scanners passed (skills.sh audits).
What problem does it solve?
You want to train or benchmark a world model but need a concise, accurate briefing on WildWorld’s scale, per-frame labels, and WildBench without rereading papers.
Who is it for?
Researchers and indie ML builders exploring action-conditioned video or game world models who need agent-ready dataset context.
Skip if: Typical SaaS shipping workflows, production observability, or builders with no generative simulation or dataset research need.
When should I use this skill?
Use WildWorld dataset, load WildWorld ARPG data, WildBench evaluation, or generative ARPG / action-conditioned video dataset work.
What do I get? / Deliverables
You and your agent share the same dataset vocabulary—frames, actions, annotations, and benchmark hooks—so prototyping and eval plans stay aligned with WildWorld.
- Structured dataset factsheet for agent context (scale, labels, entities)
- WildBench-oriented evaluation framing when benchmarking
- Consistent terminology for actions, states, and per-frame annotations
Recommended Skills
Journey fit
Dataset skills belong in Idea when you are researching what data exists before scoping a generative world-model or video prediction experiment. Research subphase captures literature-style discovery of scale, labels, and benchmark hooks (WildBench) prior to prototype design.
How it compares
Use as dataset and benchmark orientation, not as a training pipeline installer or hosted data API integration.
Common Questions / FAQ
Who is wildworld-dataset for?
ML-focused solo builders and agents working on generative world models, ARPG simulation, or WildBench evaluations.
When should I use wildworld-dataset?
During Idea research when comparing action-conditioned datasets, and again in Validate when scoping a prototype that needs WildWorld annotations or WildBench metrics.
Is wildworld-dataset safe to install?
The skill is documentation-heavy; review the Security Audits panel on this page and follow official dataset access terms when downloading assets.
SKILL.md
READMESKILL.md - Wildworld Dataset
# WildWorld Dataset Skill > Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection. ## What WildWorld Is **WildWorld** is a large-scale action-conditioned world modeling dataset automatically collected from a photorealistic AAA action role-playing game (ARPG). It is designed for training and evaluating **dynamic world models** — generative models that predict future game states given past observations and player actions. ### Key Statistics | Property | Value | |---|---| | Total frames | 108M+ | | Actions | 450+ semantically meaningful | | Monster species | 29 | | Player characters | 4 | | Weapon types | 4 | | Distinct stages | 5 | | Max clip length | 30+ minutes continuous | ### Per-Frame Annotations Every frame includes: - **Character skeletons** — joint positions for player and monsters - **Actions & states** — HP, animation state, stamina, etc. - **Camera poses** — position, rotation, field of view - **Depth maps** — monocular depth for each frame - **Hierarchical captions** — action-level and sample-level natural language descriptions --- ## Project Status > ⚠️ As of March 2026, the dataset and WildBench benchmark have **not yet been released**. Monitor the repository for updates. ```bash # Watch the repository for dataset release # https://github.com/ShandaAI/WildWorld ``` --- ## Repository Setup ```bash # Clone the repository git clone https://github.com/ShandaAI/WildWorld.git cd WildWorld # Install dependencies (when benchmark code is released) pip install -r requirements.txt ``` --- ## Expected Dataset Structure Based on the paper and framework description, the dataset is expected to follow this structure: ``` WildWorld/ ├── data/ │ ├── sequences/ │ │ ├── stage_01/ │ │ │ ├── clip_000001/ │ │ │ │ ├── frames/ # RGB frames (e.g., PNG) │ │ │ │ ├── depth/ # Depth maps │ │ │ │ ├── skeleton/ # Per-frame skeleton JSON │ │ │ │ ├── states/ # HP, animation, stamina JSON │ │ │ │ ├── camera/ # Camera pose JSON │ │ │ │ └── actions/ # Action label files │ │ │ └── clip_000002/ │ │ └── stage_02/ │ └── captions/ │ ├── action_level/ # Per-action descriptions │ └── sample_level/ # Clip-level descriptions ├── benchmark/ │ └── wildbench/ # WildBench evaluation code ├── assets/ │ └── framework-arxiv.png ├── LICENSE └── README.md ``` --- ## Working with the Dataset (Anticipated API) ### Loading Frame Annotations ```python import json import os from pathlib import Path from PIL import Image import numpy as np class WildWorldClip: """Helper class to load a WildWorld clip and its annotations.""" def __init__(self, clip_dir: str): self.clip_dir = Path(clip_dir) self.frames_dir = self.clip_dir / "frames" self.depth_dir = self.clip_dir / "depth" self.skeleton_dir = self.clip_dir / "skeleton" self.states_dir = self.clip_dir / "states" self.camera_dir = self.clip_dir / "camera" self.actions_dir = self.clip_dir / "actions" def get_frame(self, frame_id: int) -> Image.Image: frame_path = self.frames_dir / f"{frame_id:06d}.png" return Image.open(frame_path) def get_depth(self, frame_id: int) -> np.ndarray: depth_path = self.depth_dir / f"{frame_id:06d}.npy" return np.load(depth_path) def get_skeleton(sel