Zero-Server Processing: Your images and JSON never leave your device. All logic is executed client-side via your browser's RAM.
1. Upload KoboldAI JSON
Technical Documentation: Mapping Legacy Backend Saves to Classical Prefixes
The migration from **KoboldAI** to **Pygmalion** JSON represents a transition between two of the AI roleplay community's most important foundational data models. While KoboldAI's format is designed primarily to preserve the context state of an inference session, the Pygmalion standard serves as a generalized, prefix-heavy bridge for classic frontends and early inference tools.
1. Semantic Field Normalization
Pygmalion characters utilize explicit key prefixes like char_name, char_persona, and world_scenario. KoboldAI JSON files (typically found in legacy Lite frontends) utilize a more traditional flat structure with keys like description and scenario. Our converter identifies these backend-focused keys and performs a "Semantic Alignment," ensuring that the character's internal logic and behavioral goals are correctly mapped to the foundational Pygmalion blocks.
2. Dialogue and Example Sanitization
Classic Pygmalion-compatible models perform best with clean, narrative-driven dialogue examples. KoboldAI save files are often formatted for raw text completion, which can include messy tokens or inconsistent formatting. Our conversion engine identifies and reformats dialogue within the mes_example field, ensuring it adheres to the standardized patterns that maximize character "soul" without "token leaking" legacy artifacts that might trigger formatting errors in classic roleplay interfaces.
3. Context and Identity Preservation
Both KoboldAI and Pygmalion users rely heavily on the character's backstory and scenario. Our tool ensures that the core narrative intent of your Kobold characters is meticulously preserved in the world_scenario and char_persona blocks of the new Pygmalion JSON. This provides classic inference tools with the best possible roadmap for generating relevant, character-aligned responses that stay true to the original creator's vision.
4. Privacy and the "Zero-Server" Mandate
Just like the foundational tools we bridge, our converter prioritizes privacy. By using a "Zero-Server" architecture, we perform the entire schema translation within your browser's RAM via JavaScript. Your character files, intimate details, and creative backstories are never transmitted across the open web. This ensures that you maintain absolute control over your digital assets as you move them between the industry's most established local AI standards.
5. Future-Proofing Legacy Characters
By outputting a standardized Pygmalion JSON, we ensure that your KoboldAI characters remain compatible with a vast array of legacy tools and specialized roleplay frontends. This conversion process isn't just about moving a file; it's about preserving your creative history and ensuring your digital characters remain accessible, editable, and compatible with the broader AI ecosystem, regardless of the evolving landscape of inference technology.
