← Back to Plugin Center

CommunityVisionby liustack★ 3.7k

modlens

The first vision plugin for dsh: paste an image, get structured evidence back.

npx -y @deepseek-ai/dsh plugin --profile web add @liustack/modsearch@latest

GitHub Repo ↗

The first vision plugin for DeepSeek Harness and the vision bridge for every text-only coding agent: paste an image and instantly get structured JSON evidence (OCR, layout, chart data), so text-only models can finally "see".

It injects image understanding results as evidence into context, so models without multimodal capabilities can reason over screenshots — error dialogs, design mockups, and table photos all just work.

Features

• Paste an image, get structured JSON evidence
• OCR, layout, and chart data in one pass
• Gives text-only models eyes
• The most-starred vision plugin

Browse all