Skip to content
Finalist

yolo dex

We built an end-to-end “video → dataset → agent” pipeline as a Codex Skill plugin that can turn any YouTube video into a training-ready vision dataset.

Code Video Cerebral Valley

Core trick

Downloads YouTube frames, fans them out to parallel Codex agents that draw bounding boxes, and exports a YOLO-format dataset for training.

Tags

visionagents

Description

We built an end-to-end “video → dataset → agent” pipeline as a Codex Skill plugin that can turn any YouTube video into a training-ready vision dataset. You paste a URL, and the skill runs locally to download the video, extract frames, and spin up as many parallel Codex labeling agents as needed to generate bounding boxes for the objects of interest. It exports a full YOLO dataset on-device, which can immediately train a detector. That detector becomes a game-state extractor from pixels, enabling bots to play any game without emulator hooks or engine access, essentially compressing dataset creation + model bootstrapping into one repeatable command.
qtzx06/yolodex★ 9613 forksPythonCreated Feb 5, 2026 · last push Apr 8, 2026

README excerpt

<h1 align="center">yolodex</h1>

<div align="center">

![GitHub last commit](https://img.shields.io/github/last-commit/qtzx06/yolodex?style=for-the-badge&labelColor=ffffff&color=8ecaff)
![GitHub Repo stars](https://img.shields.io/github/stars/qtzx06/yolodex?style=for-the-badge&labelColor=ffffff&color=8ecaff)
![GitHub repo size](https://img.shields.io/github/repo-size/qtzx06/yolodex?style=for-the-badge&labelColor=ffffff&color=8ecaff)
![License: MIT](https://img.shields.io/badge/license-MIT-8ecaff?style=for-the-badge&labelColor=ffffff)
![PRs welcome](https://img.shields.io/badge/PRs-welcome-8ecaff?style=for-the-badge&labelColor=ffffff)

</div>

<div align="center">

codex-native yolo dataset + training agent skills for any unlabeled data. codex skills help determine how to properly label and splice your data.

</div>

[![star history chart](https://api.star-history.com/svg?repos=qtzx06/yolodex&type=Date)](https://star-history.com/#qtzx06/yolodex&Date)

## architecture

```mermaid
flowchart LR
  A["data source (any unlabeled data)"] --> B["collect"]
  B --> C["label"]
  C --> D["augment"]
  D --> E["train"]
  E --> F["eval"]
  C -. "parallel subagents in git worktrees" .-> C
  F --> G{"meets target?"}
  G -- "no" --> C
  G -- "yes" --> H["complete"]
```

## components

- skills runtime: [codex skills](https://developers.openai.com/codex/skills/)
- pipeline skills: `collect`, `label`, `augment`, `train`, `eval`, `yolodex`
- orchestration loop: `yolodex.sh` + `AGENTS.md`
- shared

Team