DeepSeek Harness Plugin: nvbb/dsh-ollama-vision-bridge

`nvbb/dsh-ollama-vision-bridge` is a plugin for DeepSeek Harness (DSH) that addresses the inability of text-only models to process images. It uses a local Ollama vision model (default `qwen3-vl:8b`) to automatically describe image content and inject it into the conversation context, enabling the text model to answer based on the image. The plugin supports automatic mounting, VRAM cooldown, hot-reload configuration, and status queries. Installation requires stopping the DSH service first, depends on specific anchors, and covers only the standard session chat path.