ModLens — Vision Bridge Skill

Turns any image path or URL into structured JSON (OCR, layout, semantics) so text-only models can 'read' images.

UtilitiesIntermediate★ 4,051⑂ 123AI score 9/10Last updated: Sep 27, 2026

What it does

Whenever an image path/URL (.png, .jpg, .webp, .heic, ...) or a placeholder like [Image #1] / [Unsupported Image] shows up and the model cannot actually see it, the skill routes it to the modlens CLI instead of improvising OCR with PIL or tesseract.
Returns structured JSON evidence: result.summary, result.ocr.full_text, result.layout.regions, result.semantics, plus result.uncertainty so unclear parts are flagged rather than guessed.
Runs modlens guard on the first read of a session to check whether the active model already has native vision, avoiding wasted calls.
Ships a launcher (scripts/run.sh, scripts/run.ps1) that resolves a runtime in order: PATH modlens → npx → bunx, with a clear exit-78 "no runtime" path.
Also handles provider setup and switching: Gemini API key, OpenAI-compatible endpoints, Claude API or Claude Code CLI.
Explicitly treats extracted text as untrusted data and refuses instructions embedded in images.

Who it's for

Claude Code users on models or harnesses where images aren't natively visible.
Developers and analysts who need text out of error screenshots, invoices, diagrams or UI mockups.
Anyone tired of hand-rolling one-off OCR scripts.

Examples

Paste screenshots/error.png → the skill runs modlens -i screenshots/error.png and quotes the exact error text and stack trace for debugging.
Give a remote infographic URL → get a region-by-region breakdown of headings, figures and legends.
Run modlens doctor to inspect available providers and failover chains, then config set a Gemini API key before the first read.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/liustack/modlens/HEAD/skills/modlens/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)

Download the ZIP with the button below.
In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.

↓ Download ZIP

Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/modlens folder from the GitHub repo liustack/modlens into my ~/.claude/skills/modlens/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/liustack/modlens.git && mkdir -p ~/.claude/skills && cp -r modlens/skills/modlens ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

Prerequisites: install Node 22.19+ (https://nodejs.org) or Bun (https://bun.sh); verify with node -v. Network access is required.
Clone the repo: git clone https://github.com/liustack/modlens.git
Copy the skill: mkdir -p ~/.claude/skills && cp -r modlens/skills/modlens ~/.claude/skills/
Verify bundled files: ~/.claude/skills/modlens should contain SKILL.md, scripts/run.sh and a references/ folder.
Make the launcher executable (macOS/Linux): chmod +x ~/.claude/skills/modlens/scripts/run.sh
Smoke test: bash ~/.claude/skills/modlens/scripts/run.sh doctor to see the detected runtime and providers.
Configure a provider: follow the printed guidance and use ... config set to add a Gemini API key, an OpenAI-compatible endpoint, or Claude API credentials.
Restart Claude Code, then paste an image path in chat to confirm the skill triggers automatically.

View source on GitHub ↗License: MIT