JochenYang/opencode-vision

OpenCode vision support: automatically saves pasted images and guides non-vision models to recognize them via tool calling. Supports single & multiple images.