dsh-img
Run the following command in DeepSeek Harness:
dsh plugin install gmleong/dsh-img
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install gmleong/dsh-img in DeepSeek Harness to install this plugin. Full source at https://github.com/gmleong/dsh-img
About this plugin
The biggest gap in any text-only model is its inability to look at a screenshot, a design mockup, or a crash dialog. dsh-img bridges that gap for DeepSeek Harness: it connects any text-based coding agent to a vision API for image Q&A and element grounding, and it ships a set of fully key-free local pixel tools for cropping, pixel-level diffing, dominant-color extraction, OCR, bitmap-to-SVG tracing, and background removal.
On the backend side, the plugin defaults to Zhipu GLM-4V-Flash or Tongyi Qwen-VL (both offering free tiers), or can point straight at a local Ollama vision model for a zero-key, zero-external-dependency setup. A single backends array enables ordered failover across multiple providers out of the box, and recognition results are cached by image and question hash so repeat calls cost nothing.
Built for developers who need their text agent to read UI screenshots, verify design fidelity, or vectorize icons without juggling multiple API keys. Pure JavaScript, zero build steps, one command to plug into an existing dsh workflow.
Use Cases
- Let text models read UI screenshots and error dialogs
- Pixel-level diffing for design fidelity checks
- Bitmap-to-SVG tracing and background removal
Best For
- Developers building text-based coding agents
- Solo users who want vision without juggling APIs
- Offline developers using local Ollama models
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.