Not sure how to describe the shot? The agent calls search_templates, finds a fitting prompt from the library (RU/EN), and plugs the recommended model straight into generate_image or generate_video.
search_templates is a tool of the Clipia MCP server for hybrid (text and semantic) search over a library of 3500+ ready prompts for images and video. Each result contains the prompt itself, the media type, and a recommended model — the agent can pass it straight into generate_image or generate_video. This removes the blank-page problem: instead of inventing wording, the agent finds a proven template.
| Parameter | Type | Req. | Description |
|---|---|---|---|
query | string | yes | What to search for (RU or EN) |
media_type | enum image|video | — | Limit the media type |
limit | int (1–20) | — | How many templates to return (default 8) |
{
"name": "search_templates",
"arguments": {
"query": "минималистичный постер для кофейни",
"media_type": "image",
"limit": 5
}
}{
"templates": [{
"title": "Neon coffee poster",
"prompt": "minimalist neon coffee poster, dark background, ...",
"model": "nano-banana-2",
"media_type": "image",
"category": "marketing"
}],
"note": "pass prompt + model into generate_image"
}The agent finds a proven prompt instead of guessing the wording.
Each template is already tied to a fitting model — the result is more predictable.
Hybrid search understands queries in both languages and matches by meaning, not just words.
claude mcp add --transport http clipia https://api.clipia.ai/mcp \
--header "Authorization: Bearer clipia_live_XXXX"For claude.ai and ChatGPT no key is needed — sign in with your account. Create a key in Clipia settings.
Template search spends no credits — credits are spent only on the generation itself (generate_image / generate_video).
It is a curated library of 3500+ image and video templates tied to models.
Yes, hybrid search works in Russian and English and matches by meaning.
The agent plugs prompt and model into generate_image or generate_video and starts the generation.