- Fixed issue where Qwen2.5-VL were always loaded in text-only mode even when a valid mmproj file was specified.
- Improved mmproj auto-detection logic.
Add the publish workflow to enable Comfy Registry releases.
- Trigger publishing only on changes to pyproject.toml on the main branch
- Keep manual workflow dispatch for controlled runs
- Prevent publishing from forks via repository owner check
No changes to runtime behavior or node functionality.
Restrict the publish workflow trigger to the main branch only.
- Remove the unused master branch from the trigger list
- Clarify the intended release flow and avoid accidental publishes
No functional changes to the custom nodes themselves.
Add and refine pyproject.toml metadata in preparation for Comfy Registry publishing.
- Align version to v1.0.6
- Clean up dependencies to include only actual runtime requirements
- Update project description and documentation URL
- Keep PublisherId empty for initial merge
No functional or runtime behavior changes.
- Update version to 1.0.6 to match the current release
- Clean up dependencies to include only actual package requirements
- Set Documentation URL to the main README
No functional or runtime behavior changes.
Improved stability when switching between Qwen3-VL GGUF models
Fixed mmproj reuse issues in local vision models
Refined internal GGUF model lifecycle management
Documentation updates:
- Clarified project scope as a prompt generator for QwenImageEdit and Wan2.2
- Reorganized Credits and Dependencies for clearer attribution
- Updated llama-cpp-python installation notes to reference the JamePeng fork documentation
Internal structure preparation for future backend refactoring
No breaking changes to node interfaces
- Vision LLM Node: Local GGUF vision model with multi-image support
- Qwen Image Edit Prompt Generator: Image editing optimization
- Wan Video Prompt Generator: T2V/I2V prompt enhancement
- Unified interface: CPU/GPU device selection, mmproj auto-detect
- GPL-3.0 licensed with proper attribution
Based on ComfyUI-QwenPromptRewriter (lihaoyun6) and ComfyUI-QwenVL (1038lab)
