LABSCO SUMMARY

A capability adapter, not a vision model — the quality is whatever service you point it at.

The useful design choice is vision_update_config: the key, model and endpoint are all changeable at runtime, so switching providers is a tool call rather than an edit-and-restart. vision_get_config masks the key when it shows the configuration, which matters because the model reads that output too.

What it is

An MCP server that sends images to an external, OpenAI-compatible vision service and returns a text description. It exists so a text-only model can answer questions about a picture.

What you get

vision_analyze — describe one or more images, from absolute local paths or public URLs, with a prompt of your own
Common raster formats accepted, with a per-call limit on how many images and how large each may be
vision_get_config — show the current service configuration, with the key masked
vision_update_config — change the key, the model or the service address at runtime, without restarting
Any OpenAI-compatible vision endpoint works, so the provider is your choice

Requirements

Node, and a key for a vision service. The key can be set through the configuration tool, or passed in the environment if you would rather it never touched disk. The service address and model default to one provider and can be pointed at any compatible one.

Setup effort

One command plus a key — npx -y @guorui99/vision-mcp, then supply credentials

See installation options

USE CASES

01

Asking a text-only model about a screenshot

Hand it the local path and the question; the description comes back as text.

vision_analyze

02

Comparing two images

Pass both in one call with a prompt asking what differs between them.

vision_analyze

03

Switching to a different vision provider

Change the key, model and endpoint in place instead of restarting the server.

vision_update_config

04

Checking which service is actually configured

Read the current configuration back, with the key masked.

vision_get_config

3 tools

Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.

`vision_analyze`

Analyze one or more images using a vision-capable model.

Accepts local file paths (absolute) or public URLs. Supports jpg, jpeg, png, gif, webp, bmp formats. Returns a detailed text description of image contents.

Examples:

"What's in this photo?" -> { images: ["/path/to/photo.jpg"] }
"Compare these two images" -> { images: ["img1.png", "img2.png"], prompt: "Compare these two images" }
"Read the text in this image" -> { images: ["https://example.com/screenshot.png"], prompt: "Extract all text from this image" }

Args:

images (string[]): Array of file paths or URLs (1-5 images)
prompt (string): What to look for in the image(s), defaults to "用中文详细描述..."
model (string, optional): Override the configured vision model

Returns: Detailed text description of the image(s).

Error Handling:

"文件不存在: <path>" if a local file is not found
"不支持的格式: .<ext>" for unsupported file types
"API 请求失败: ..." if the upstream vision API fails

BEFORE YOU INSTALL

What to know first

Tools you get

3 tools when we connected. vision_update_config writes your API key, model name and base URL to config.json, so you can start unconfigured. Set VISION_API_KEY instead if you want nothing written to disk.

Requires

npx on PATH
Node.js >= 18
VISION_API_KEY in your environment

Where it runs

The documented client runs the server on your machine.

CONTEXT COST

Light Standing cost

Rated by the size of the 3 tool definitions that ride along on every turn, not by how many there are

INSTALLATION

Connect your client

Server setup

TERMINAL

npx -y @guorui99/vision-mcp

MCP Clients

Claude Desktop

CLAUDE DESKTOP · JSON

{
  "mcpServers": {
    "vision": {
      "command": "npx",
      "args": [
        "-y",
        "@guorui99/vision-mcp"
      ]
    }
  }
}

Set the key after the server starts by asking your assistant to run the vision_update_config tool with api_key="your-api-key", or put VISION_API_KEY in the server's environment, which writes nothing to disk. The README's command-line example: VISION_API_KEY=your-key npx -y @guorui99/vision-mcp.

Vision MCP

A capability adapter, not a vision model — the quality is whatever service you point it at.

What it is

What you get

Requirements

Setup effort

Asking a text-only model about a screenshot

Comparing two images

Switching to a different vision provider

Checking which service is actually configured

3 tools

`vision_analyze`

`vision_get_config`

`vision_update_config`

What to know first

Tools you get

Requires

Where it runs

Light Standing cost

Connect your client

Server setup

MCP Clients

Claude Desktop

Vision MCP

A capability adapter, not a vision model — the quality is whatever service you point it at.

What it is

What you get

Requirements

Setup effort

Asking a text-only model about a screenshot

Comparing two images

Switching to a different vision provider

Checking which service is actually configured

3 tools

vision_analyze

vision_get_config

vision_update_config

What to know first

Tools you get

Requires

Where it runs

Light Standing cost

Connect your client

Server setup

MCP Clients

Claude Desktop

Same category

Moondream

MCP OpenVision

AI Image MCP Server

DINO-X MCP Server

`vision_analyze`

`vision_get_config`

`vision_update_config`