Gyazo Mcp Server

What is Gyazo Mcp Server about?

Enables AI models to interact with Gyazo‑hosted images as structured resources. Images are addressed with gyazo-mcp:// URIs and come with metadata, OCR text (when available), and raw content.

How to use Gyazo Mcp Server?

1. Install the server globally via npm:

   npm install -g @notainc/gyazo-mcp-server
   

2. Set the access token obtained from Gyazo:

   export GYAZO_ACCESS_TOKEN=your-token-here
   

3. Run the server (recommended via npx):

   npx -y @notainc/gyazo-mcp-server
   

4. Integrate with clients such as Claude Desktop by adding a configuration entry (see the README for the JSON snippet).

Key features

• Resources: Access each Gyazo image via a gyazo-mcp:// URI with original binary data, metadata (title, description, app, URL) and optional OCR text.
• Search tool (gyazo_search): Full‑text search across a user's captures, supporting keyword, title, app, URL, date range and pagination.
• Fetch tool (gyazo_image): Retrieve a specific image by ID or URL, returning both content and detailed metadata.
• Latest image tool (gyazo_latest_image): Get the most recent upload with metadata and OCR.
• Upload tool (gyazo_upload): Upload a base64‑encoded image, optionally attaching title, description, referer URL and app name; returns the new image's permalink and ID.
• TypeScript implementation with Docker support for containerised deployment.

Use cases

• Allow AI assistants to cite visual evidence from a user's Gyazo library when answering questions.
• Automate content creation pipelines that need to fetch, annotate, or reuse screenshots and screenshots‑derived graphics.
• Build custom workflows that trigger actions based on newly uploaded Gyazo images (e.g., posting to Slack, archiving).
• Enable searchable image archives for knowledge‑base bots that require visual context.

FAQ

Q: Which environment variable is required? A: GYAZO_ACCESS_TOKEN must contain a valid Gyazo API token.

Q: Can I run the server in Docker? A: Yes. Build the image with npm run image:build or use the provided Docker command in the README.

Q: What image formats are supported? A: All formats supported by Gyazo (JPEG, PNG, GIF, WebP, etc.).

Q: How does pagination work for gyazo_search? A: The tool accepts page and per_page parameters and returns a slice of matching results.

Q: Is OCR data always available? A: OCR text is included only when Gyazo has processed the image with OCR; otherwise the field is empty.

gyazo-mcp-server

A Model Context Protocol server for Gyazo image integration

This is a TypeScript-based MCP server that provides access to Gyazo images. It allows AI assistants to access and interact with Gyazo images through the Model Context Protocol, providing:

• Resources representing Gyazo images with URIs and metadata
• Tools for searching, fetching, and uploading images
• Image content and metadata access via the Gyazo API

Features

Resources

• List and access Gyazo images via gyazo-mcp:// URIs
• Each image includes:
  • Original image content
  • Metadata (title, description, app, URL)
  • OCR data (if available)
• Supports various image formats (JPEG, PNG, etc.)

Tools

• gyazo_search - Full-text search for captures uploaded by users on Gyazo

  • Search by keyword, title, app, URL, or date range
  • Supports pagination for browsing multiple results
  • Returns matching image URIs and metadata

• gyazo_image - Fetch image content and metadata from Gyazo

  • Retrieve specific images by ID or URL
  • Returns both image content and detailed metadata

• gyazo_latest_image - Fetch the most recent image from Gyazo

  • Returns both image content and metadata
  • Includes OCR text if available

• gyazo_upload - Upload an image to Gyazo

  • Upload images with base64 encoded image data
  • Add optional metadata like title, description, referer URL, and app name
  • Returns the uploaded image's permalink URL and ID

Installation

NPM Package

The easiest way to install the Gyazo MCP server is via npm:

npm install -g @notainc/gyazo-mcp-server

Prerequisites

• Create a Gyazo account if you don't have one: https://gyazo.com
• Get your Gyazo API access token from: https://gyazo.com/api
  • Click "Register applications" button
  • Click "New Application" button
  • Fill in the form with your app name and description
    • Name and Callback URL are required
    • You can use http://localhost for the Callback URL
  • Click "Submit" button
  • Click application name to view details
  • Scroll down to "Your Access Token"
  • Click "Generate" button
  • Copy "Your access token" value
• Set the GYAZO_ACCESS_TOKEN environment variable with your token

Claude Desktop Integration

To use with Claude Desktop, add the server config:

On MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json On Windows: %APPDATA%/Claude/claude_desktop_config.json

Using NPM package (recommended)

{
  "mcpServers": {
    "gyazo-mcp-server": {
      "command": "npx",
      "args": ["@notainc/gyazo-mcp-server"],
      "env": {
        "GYAZO_ACCESS_TOKEN": "your-access-token-here"
      }
    }
  }
}

Using Docker (optional)

{
  "mcpServers": {
    "gyazo-mcp-server": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "GYAZO_ACCESS_TOKEN",
        "gyazo-mcp-server"
      ],
      "env": {
        "GYAZO_ACCESS_TOKEN": "your-access-token-here"
      }
    }
  }
}

Development

Install dependencies:

npm ci

Build the server:

npm run build

For development with auto-rebuild:

npm run watch

Docker Build (optional)

npm run image:build

---