gyazo

This Gyazo MCP server provides a Model Context Protocol interface to access Gyazo images. It exposes resources representing Gyazo captures with URIs and rich metadata, along with tools to search, fetch, and upload images via the Gyazo API. You can query images by keywords or metadata, retrieve image content and OCR (if available), and upload new images with optional title, description, referer URL, and app name. The server is designed to be consumed by AI assistants or client tools that integrate with the MCP to present image assets and related information to users.

To use the server, configure a client or integration to connect to the MCP endpoint exposed by your deployment. The primary tools available are gyazo_search for full-text search across user uploads, gyazo_image for fetching image content and metadata by ID or URL, gyazo_latest_image for the most recent Gyazo capture (with OCR when present), and gyazo_upload for uploading new images with metadata. Ensure you supply your Gyazo access token via the GYAZO_ACCESS_TOKEN environment variable, as this is required for authenticated API access. Once configured, you can browse results, request specific images, or push new images into Gyazo through the MCP tools.

Prerequisites:

• Node.js (recommended latest LTS)
• npm or pnpm
• A Gyazo account with an access token

Installation steps:

1. Install the MCP server globally via npm:

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

2. Set up your environment with your Gyazo access token:

export GYAZO_ACCESS_TOKEN=your-access-token-here

3. Run the MCP server using npx (as recommended by the MCP docs):

npx -y @notainc/gyazo-mcp-server

4. If you prefer Docker, you can run an image (adjust token env as needed):

docker run -i --rm -e GYAZO_ACCESS_TOKEN=your-access-token-here gyazo-mcp-server

5. If you deploy to Claude Desktop or another client, configure the MCP server in your client’s settings with the appropriate command and environment variables as shown in the examples.

Note: You may also use the npm package approach in your environment’s configuration, e.g., via Claude’s config.json, as an alternative to running npx directly.

Environment variables and config tips:

• GYAZO_ACCESS_TOKEN is required to access Gyazo’s API. Keep this token secure and avoid committing it to version control.
• When using Docker, you can pass the token through -e GYAZO_ACCESS_TOKEN to the container.
• The gyazo_search tool supports keyword, title, app, URL, and date range queries with pagination; use it to navigate large image collections.
• The gyazo_upload tool can accept base64-encoded image data along with optional metadata (title, description, referer URL, app). Ensure the payload meets Gyazo’s API requirements.
• If OCR data is not available for a given image, the response will omit OCR fields gracefully.
• For Claude Desktop integration, ensure the claude_desktop_config.json path is correct for your OS and that the server is reachable from Claude.