I found an interesting way to avoid making a single model do all the coding work
Fable-orchestrator — ★525 It’s a skill for Codex that splits the task into three roles: ▸ Fable plans the work and makes the final decision ▸ GPT-5.6 Luna writes the standard implementation ▸
GPT-5.6 Luna + Fable 5.1 working together on your coding tasks 😳
Fable Orchestrator lets Codex route different parts of a task to specialized agents getting started: 1. open 2. install the Fable Orchestrator skill 3. open it in Codex 4. run /fable
Using DeepSeek Harness, I created a Mario-style browser game by asking AI to share the roles 🎮
🧠 Codex: Planning/Review 🛠 Qwen3.8-Flash-Next NVFP4: Locally implemented Qwen fixed the screen display corruption found by Codex. Passed 23 internal tests and 49 browser verifications, and passed the final review!
AI Takes Over Wei, Shu, and Wu in New Three Kingdoms
Netizens simulated using Doubao, Qianwen, and Deepseek to take over the Wei, Shu, and Wu kingdoms during the Three Kingdoms period. The final result was to see who emerged victorious.
You have reached the end of the archive
All of deepseek