Purpose: Pexafy MCP is a remote MCP server that lets AI assistants search directly for real photos available for free use during a conversation. Users can describe the scene requirements as they would to a colleague, and the assistant generates the search; Pexafy matches scenes semantically rather than by tags.
Main capabilities: Supports retrieval using natural language descriptions, example images, or similarity; can return sixteen photos at a time, supports zooming in to view, swiping to browse, saving favorites, and using "≈" to find similar images; selected results are returned to the assistant in the user's order; users can also specify in text, such as "the third one, but wider." Selected photos can continue to be used by the assistant for articles, slides, landing pages, posts, etc. Each photo retains its license and the attribution line that must be displayed, and is provided as an image file when needed. An optional intranet grid interface is also provided.
How to use: In editors such as VS Code and Cursor, you can search directly, with no installation, no need to paste an API Key, and no account required; ChatGPT and Claude require logging in to Pexafy once, and a free account is sufficient. Five tools are provided, all read-only.
Photo sources: Pexafy searches its own index of millions of free photos, with material from image libraries such as Unsplash, Pexels, and Pixabay; each photo is indexed by meaning, so a complete scene description in one sentence is better than listing keywords. Each result indicates the source, license, and the attribution line that must be displayed.
Limitations: The materials do not provide specific installation commands, pricing plans and prices, a complete list of supported platforms, index update frequency, or performance data, and none are currently provided.
