ai-wdywfm is a prompt assistant built specifically for Stable Diffusion WebUI Forge Neo. It helps beginners describe what they want without first learning prompt syntax, while giving experienced users a faster way to draft model-aware prompts.
More info here -> https://github.com/mikhailfur/sd-webui-ai-wdywfm
Description
Changelog
[v0.1-stable] - 2026-07-27
🎉 First Stable Release
Initial stable release of AI WDYWFM for Stable Diffusion WebUI Forge Neo.
✨ Added
- Initial integration with Stable Diffusion WebUI Forge Neo.
- Support for OpenRouter API.
- Support for LM Studio and OpenAI-compatible local servers.
- Natural language → Stable Diffusion prompt generation.
- Optional image input for vision-capable LLMs.
- Automatic detection of installed Checkpoints and LoRAs.
- LoRA metadata and trigger-word extraction.
- JSON Schema validation with automatic response repair.
- Prompt dialect support (Natural / Booru / Auto).
- Read-only generation recommendations (CFG, Steps, Sampler, Size).
- Secure Windows DPAPI encryption for stored API keys.
- Built-in diagnostics viewer and sanitized rotating logs.
- Model discovery and selection.
- Gemma 4 compatibility improvements.
- Localization of documentation (EN, RU, KO, JA, ZH-CN, ZH-TW).
🔒 Security
- API keys encrypted using Windows DPAPI.
- Sensitive information excluded from diagnostic logs.
- Validation of generated checkpoint and LoRA identifiers before applying prompts.
⚡ Performance
- Cached LoRA metadata and sidecar parsing.
- Reduced prompt context size while preserving model accuracy.
- Optimized structured output handling.
🧪 Testing
- Added automated tests for:
- provider integrations;
- prompt validation;
- diagnostics;
- UI behavior;
- checkpoint and LoRA inventory;
- response repair.
📚 Documentation
- Added installation guide.
- Added architecture documentation.
- Added JSON Schema specification.
- Added roadmap.
- Added multilingual documentation.
📄 License
- Licensed under GNU GPL v3.