Your LLM’s wardrobe of 100+ alternate identities

What it does This repository is essentially a wardrobe for large language models: a curated collection of over 100 “skill” files—structured system prompts—that impose specific personas, writing styles, and decision-making frameworks onto agents. The files cover everything from classical Chinese scholars and literary novelists to social-media influencers and emotional-support companions, all formatted for import into compatible agent runtimes like OpenClaw and StepClaw. Several folders go further by bundling multiple personas into collaborative teams, such as a council of twenty legendary Tianya forum veterans or a writers’ room of 31 famous film directors.

The interesting bit The project treats prompt engineering less like instruction tuning and more like method acting: each skill attempts to capture a figure’s “narrative DNA” or “visual language” through handcrafted system prompts rather than fine-tuning weights. The most unusual experiments are the multi-agent ensembles—turning a bulletin-board’s worth of internet folklore or a filmography into a debating committee or sequential chain of critics.

Key highlights

• Over 100 standalone persona skills, from workplace coaches to anime characters, stored as markdown files.
• Multi-agent “teams” including a 31-director creative consortium and a 20-member “Tianya gods” advisory council.
• Specialized distillations for literary writers, classical texts (e.g., Yongle Encyclopedia), and Chinese social-media influencers.
• Compatibility badges for OpenClaw, StepClaw, and nuwa-skill ecosystems.
• Includes a PDF实操指南 (practical operation guide) for skill distillation.

Caveats

• The README is heavy on emoji and superlatives (“全网最全”, “最火爆”) but light on technical validation of how faithfully these prompts reproduce the claimed personas.
• Most content is in Chinese, and the “distillation” appears to be manual prompt authoring rather than an automated or algorithmic process, which may not match expectations set by the term “distillation.”

Verdict Grab this if you are building Chinese-language agent applications and need ready-made character prompts or creative multi-agent scenarios without writing them from scratch. Skip it if you are looking for a training framework, automated persona extraction pipeline, or English-first content.