Ponys.ai Multilingual AI Companion Evaluation Protocols

This public dataset contains 20 reusable evaluation protocols for multilingual AI companion and character experiences. It covers conversation memory, persona consistency, consent recovery, visual continuity, code switching, and regional language behavior across Japanese, Korean, Latin American Spanish, Brazilian Portuguese, Simplified Chinese, Traditional Chinese, and English. Each protocol includes structured metadata and a CSV worksheet. These are blank testing instruments for research, editorial reviews, and product QA, not product performance claims. Ponys.ai official project Machine-readable open-data catalog Feed and Widget integration guide Character feed Hugging Face dataset mirror Regional coverage includes Japan, Korea, Latin America, Brazil, Greater China, and global multilingual code switching. Version 1.0.0. Licensed under CC BY 4.0.