vision-mcp-server

Vision MCP Server

vision-mcp-server gives text-only AI clients access to image analysis through ModelScope vision models. Users can submit a local image or an image URL, optionally ask a specific question, and receive a detailed text description.

Server facts

vision-mcp-server

Project

License
Not specified

Tech

Runtime / language
Not specified
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
11
Last commit
Data as of

Tags

  • image analysis
  • mcp
  • vision

Overview

Image analysis in a text-only AI client is the central workflow for vision-mcp-server. Its analyze_image tool accepts an image location, either a local file path or an online URL, and returns a detailed description in text. An optional prompt lets the caller ask about objects in the image or specify another analysis request. Without a custom prompt, the tool asks for a description of the image. This supports users working with visual references and interface analysis through an assistant whose own model accepts only text.

The analysis uses vision models available through ModelScope. Configuration requires a ModelScope API token, while model selection is optional: Qwen/Qwen3-VL-30B-A3B-Instruct is the default, and another supported vision model can be selected. The project exposes its tool over MCP using stdio and includes TypeScript type definitions. Deployment options cover running the package through npx, installing it globally, or installing it locally and launching its entry point with Node.

vision-mcp-server setup notes

  • vision-mcp-server records stdio only; confirm a network transport upstream before planning remote access.
  • vision-mcp-server has no recorded license; confirm reuse terms in its repository.

Related MCP servers