# liustack/modlens

The first vision plugin for DeepSeek Harness, and the vision bridge for every text-only coding agent. Paste an image, get structured JSON evidence (OCR, layout, semantics). | 全网最强 DeepSeek Harness 外挂视觉插件，为 DeepSeek、GLM 等纯文本模型外挂视觉能力，粘贴图片即得结构化 JSON 证据（OCR、版面、语义）。

Repository: https://github.com/liustack/modlens
Canonical: https://ross.abutalabs.com/products/modlens
Homepage: https://liustack.dev
Language: TypeScript
License: MIT
License Family: permissive
Topics: agent-skills, claude-code, claude-skills, codex, deepseek, dsh-plugin, glm, harness, harness-engineering, hermes-agent, openclaw, pi-agent, text-only-llm, vision-transformer, image-to-text, multimodal, ocr, vision, cordis, dsh
Last push: 2026-08-25T12:12:27+00:00

## Health v2 (maintenance only)
Score: 78/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 13
- inputs: {"age_days": 192, "days_push": 8, "days_rel": 8, "gap_med": 0, "n_releases_24m": 78}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3700, forks 106 (observed 2026-08-28T04:08:14.791796+00:00)

## What it is
ModLens is a vision plugin for DeepSeek Harness (dsh) and other text-only coding agents that converts pasted images into structured JSON evidence including OCR, layout, and semantic analysis. It acts as a vision bridge, letting text-only models like DeepSeek and GLM read images directly from the chat.

## Use cases
- give a text-only coding agent the ability to read pasted screenshots
- extract OCR text and layout from images as structured JSON
- let DeepSeek or GLM analyze UI mockups and diagrams
- add vision capabilities to a text-only LLM workflow
- read images pasted into chat without saving files first
- convert screenshots into semantic evidence for coding agents

## When to choose
- you use DeepSeek, GLM, or another text-only model in a harness like dsh and need image understanding
- you want structured JSON output (OCR, layout, semantics) from images inside a coding agent
- you want zero-friction image input by pasting directly into chat

## When to avoid
- your model already has native multimodal vision capabilities
- you need general-purpose standalone OCR outside an agent harness
- you don't use a supported harness such as DeepSeek Harness, Claude Code, or Codex

## Facets
- artifact type: plugin
- maturity: active
- function: ocr, image-processing, computer-vision, agent-framework, mcp
- domain: artificial-intelligence, image-processing, developer-tools
- platform: cli, cross-platform
- tags: vision-plugin, deepseek-harness, text-only-llm, image-to-, agent-skills, claude-code, multimodal-bridge, ai-agents, natural-language-processing, nodejs

## Member repositories
- liustack/modlens (main) score 78

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:14.791796+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:31:03.043097+00:00, confidence not recorded.
  - readme: https://github.com/liustack/modlens (fetched 2026-08-28T04:08:14.791796+00:00, sha a35f5bcd92e6)
  - homepage: https://liustack.dev (fetched 2026-08-29T09:24:52.741851+00:00, sha 79170bf4cd60)
- Data as of 2026-08-30T08:39:29.467469+00:00.