Macos Vision MCP

Your documents and your screen stay on the machine. Only extracted text and structured JSON ever reach the model. Its 13 documented tools cover detect, image, document, screen. It runs locally over stdio via the published package.

People connecting media tools to Claude, Cursor, VS Code, or another MCP client. The project is written in JavaScript.

VERIFIED ACTIVE

LAST COMMIT 2026-08-24 · ★ 5 · #73 OF 155 MAINTAINED MEDIA · VERIFIED 2026-08-25

MIT · JavaScript servers · how we verify → /methodology

01 · Install Macos Vision MCP

before you install - you'll need

The README does not document required environment variables for a basic install.

Claude Code

claude mcp add woladi-macos-vision-mcp -- npx -y macos-vision-mcp

Claude Desktop / Cursor / VS Code - add to config

{
  "mcpServers": {
    "woladi-macos-vision-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "macos-vision-mcp"
      ]
    }
  }
}

Same JSON for Cursor. For VS Code, rename the top-level key from `mcpServers` to `servers`.

Using another client? Same JSON, different key

Claude Desktop · mcpServers

Cursor · mcpServers

VS Code · servers

Windsurf · mcpServers

Zed · context_servers

Cline · mcpServers

Roo Code · mcpServers

Continue · mcpServers

LibreChat · mcpServers

Gemini CLI · mcpServers

Codex CLI · mcp_servers

Full setup guides: every client.

02 · Evidence

Security posture

What to check before giving this server access to your agent - from the registry, GitHub, and our own probes. We don't score safety; we show what's verifiable.

runs as local process (stdio) - runs on your machine with your user's permissions

license MIT - declared in the repository

npm package macos-vision-mcp - unscoped; check the name against the project README before installing

registry namespace io.github.woladi is GitHub-verified and matches the repo owner

03 · What Macos Vision MCP can do

Prose above is summarized from the project's README and registry record - no invented capabilities.

What you can build

With this server connected, an agent can detect human faces and return their count and positions, detect the four corner points of a document in a photo (paper, receipt, ID). Useful as a crop / deskew hint before OCR, classify image content into 1000+ categories with confidence scores, and list on-screen windows with global screen-point bounds, front-to-back.

Capability map

Tools grouped from the project's README - what Macos Vision MCP lets an agent do.

Detect

3 tools - e.g. Detect human faces and return their count and positions.; Read QR codes, EAN, UPC, Code128, PDF417, Aztec, and other 1D/2D codes.; Detect the four corner points of a document in a photo (paper, receipt…

detect_faces · detect_barcodes · detect_document

read & search

3 tools - e.g. List on-screen windows with global screen-point bounds, front-to-back.; Capture + OCR in one step - read what an app shows right now, fully offline.; Find a UI element by visible text; returns clickPoint…

list_windows · read_screen_text · find_element

Ocr

1 tool - e.g. Extract text from an image or PDF (JPG, PNG, HEIC, TIFF, PDF). Returns plain text, or per-page paragraphs + text blocks with lineId / paragraphId and bounding b

ocr_image

Classify

1 tool - e.g. Classify image content into 1000+ categories with confidence scores.

classify_image

Analyze

1 tool - e.g. Returns structured JSON with reading-order paragraphs, raw text blocks (bbox / confidence), faces, barcodes, and rectangles - ready for the model to reconstruct

analyze_document

Capture

1 tool - e.g. Screenshot the main display, a window (even occluded), an app's frontmost window, or a region. Returns the file path + screen-point frame - never the image byte

capture_screen

Assert

1 tool - e.g. Local pass/fail assertion that text is present on / absent from the screen - the verdict is computed on your Mac, not by a cloud model.

assert_text

Vision

1 tool - e.g. Report macOS version, Screen Recording / Accessibility permission state, and displays.

vision_capabilities

Latest releases

v0.6.7 · 2026-08-24

a logo - viewfinder brackets locked onto an eye · Violet brackets, a red iris, a black pupil with a crosshair on it. The mark is · authored as SVG rather than kept as the generated raster: the image model returns…

v0.6.6 · 2026-08-24

keywords and descriptions that mention the half of this tool they were omitting · The npm keywords were entirely OCR and documents - nothing about screen reading, · UI work, accessibility or agents - and both one-line…

v0.6.5 · 2026-08-24

add a "What agents use this for" chapter · The README framed everything as UI testing, which undersold the tool. Most work · an agent does on a Mac needs no understanding of a layout at all - it needs to · see the…

04 · Who maintains Macos Vision MCP

macos-vision-mcp is maintained by woladi. We track 2 MCP servers from woladi - 2 actively maintained, 5 combined GitHub stars, oldest repo from Apr 2026.

05 · Facts

category
media - ranked #73 of 155 actively-maintained media servers as of 2026-08-25.
release cadence
10+ releases in the last 90 days (latest 2026-08-24)
registry
io.github.woladi/macos-vision-mcp (active, first published 2026-04-10 · 13 versions)
packages
npm:macos-vision-mcp

06 · Macos Vision MCP FAQ

Is Macos Vision MCP still maintained?

Yes - as of 2026-08-25, its last commit was 2026-08-24 and it shipped 10+ releases in the last 90 days. We re-verify nightly.

What can Macos Vision MCP do?

With this server connected, an agent can detect human faces and return their count and positions, detect the four corner points of a document in a photo (paper, receipt, ID). Useful as a crop / deskew hint before OCR, classify image content into 1000+ categories with confidence scores, and list on-screen windows with global screen-point bounds, front-to-back.

How do I install Macos Vision MCP?

Run `npx -y macos-vision-mcp`. The README does not document required environment variables for a basic install. You can also paste the ready-made client config above.

Does Macos Vision MCP run locally?

Yes - it's a stdio server: it runs on your machine (via npx) with your user's permissions. Your data stays local unless the server itself calls external APIs.

07 · Alternatives to Macos Vision MCP