The Imageory MCP server

Connect

In Claude Code, one command:

claude mcp add --transport http imageory https://imageory.in/api/mcp

Any other MCP client — add this to its config file:

{
  "mcpServers": {
    "imageory": {
      "type": "http",
      "url": "https://imageory.in/api/mcp"
    }
  }
}

Tools

• search_prompts

  Search the Imageory library of curated AI image prompts by free-text query, style, or tags. Returns ranked matches with a prompt excerpt. Use get_prompt to retrieve a full prompt once you have picked a slug.

• pick_image_for_project

  Given a description of the project you are building (its subject, audience, and visual tone), return the Imageory images that fit it best, each with a short reason. Use this when you need artwork for a site or app and do not already know what to search for.

• get_prompt

  Fetch one prompt in full by its slug, including the complete prompt text, the model it was written for, and the rendered image URL.

• adapt_prompt

  Fetch a prompt together with a structured brief for rewriting it to fit your project. This tool does not call a model — it returns the source prompt, its reusable style clauses, and explicit instructions. YOU rewrite the prompt from that material, since you already hold the project context.

• list_styles

  List the style and tag vocabulary used in the library, with counts, so you can filter searches with values that actually exist.

Rate limits

Anonymous access is fine for trying it out and for everyday use — the whole free library is public. If you hit the limit, create an API key and send it as a bearer token for a higher one.

{
  "mcpServers": {
    "imageory": {
      "type": "http",
      "url": "https://imageory.in/api/mcp",
      "headers": { "Authorization": "Bearer imk_your_key_here" }
    }
  }
}