MCP

Find templates by text, kind or tag.

Searches names, tags and descriptions across the whole catalog. Every hit carries the ids get_template and caption_meme want, plus an editor link a human can open.

It matches words, not ideas. When the user has a situation rather than a template in mind, start with suggest_memes instead — it reads the catalog and writes the captions in the same call.

Arguments

• query — string, optional. Free text, up to 200 characters. Omit it to list the most popular templates.
• kind — 'still' | 'gif', optional. Restrict to stills or looping GIFs.
• tag — string, optional. A tag slug from list_tags.
• limit — number, optional. 1 to 50. Defaults to 12.

Returns

• total — how many templates matched, before limit was applied.
• templates[] — id, slug, kind, name, tags, imageUrl and editorUrl.

Example

{
  "name": "search_templates",
  "arguments": {
    "query": "drake",
    "kind": "still",
    "limit": 2
  }
}

The structured result:

{
  "total": 7,
  "templates": [
    {
      "id": 42,
      "slug": "drake-hotline-bling",
      "kind": "still",
      "name": "Drake Hotline Bling",
      "tags": [
        "reaction",
        "preference"
      ],
      "imageUrl": "https://memeworld.afterlume.workers.dev/cdn/templates/drake-hotline-bling.jpg",
      "editorUrl": "https://memeworld.afterlume.workers.dev/meme/drake-hotline-bling-42"
    }
  ]
}

Notes

Search runs against an in-memory index of the catalog, so it is cheap to call repeatedly and safe to call before you know what you want.