liustack/modlens
The first vision plugin for DeepSeek Harness, and the vision bridge for every text-only coding agent
ModLens is a vision engine plugin that gives text-only DeepSeek and GLM models the ability to read images pasted directly into the chat. No need to save an image to a file and pass a path — just paste. Two activation modes: paste mode (image lands as a temp-file path in the composer, picked up by the modlens_read_image tool) or model-picker mode (choose a modlens vision entry from the model selector). Self-described as the most capable vision plugin for DeepSeek Harness.
Install
dsh plugin --profile web add @liustack/modlensFor a pinned version: npx -y @deepseek-ai/dsh plugin --profile web add @liustack/modlens@3.22.1. Also available as part of AIManager (desktop wrapper, zero-config). Companion web search plugin: liustack/modsearch.
Compatibility
Designed for text-only DSH models (DeepSeek, GLM). Works in the dsh web profile. Requires Node.js.
Details
- Repo: liustack/modlens
- Category: Web UI Enhancements
- Stars: 3393
- Version: 3.22.1 (latest stable at time of enrichment)
- Last push: 2026-09-18
- First seen: 2026-02-22
Recent updates
See npm @liustack/modlens for latest version. Companion plugin ModSearch provides free web search for DSH.
FAQ
- How do I install ModLens?
- Run: dsh plugin --profile web add @liustack/modlens — then restart dsh web. You can also pin a version: dsh plugin --profile web add @liustack/modlens@3.22.1
- Does ModLens work with models that already support vision?
- ModLens is designed for text-only models. If your model already has native vision, ModLens adds paste-to-path convenience but its main value is unlocking vision on text-only models.
- What image formats does ModLens support?
- ModLens handles images pasted or dropped into the composer. Common formats (PNG, JPEG, WebP, GIF) work through the paste flow.
Alternatives
Anionex/agent-vision-toolkit · omdsh-dev/dsh-genui · Anionex/dsh-vision-toolkit