Describe the clip — the agent starts generate_video, gets a request_id and the cost, then waits for the result via wait_generation. A proper long-poll, not a wrapper that times out.
generate_video is a tool of the Clipia MCP server for text-to-video. The agent sets a prompt, duration, aspect ratio, and resolution; the tool returns a request_id and the credit cost right away, after which the status is awaited with a separate wait_generation call (long-poll up to 30 seconds at a time). This is sturdier than community wrappers that drop the connection on long videos.
| Parameter | Type | Req. | Description |
|---|---|---|---|
prompt | string | yes | Clip description (EN recommended) |
model | string | — | Video model slug |
duration | number (сек) | — | Duration, usually 5–10 s |
aspect_ratio | string | — | 16:9, 9:16, 1:1 |
resolution | string | — | 480p / 720p / 1080p |
image_url | string | — | HTTPS frame → switches to image-to-video |
generate_audio | bool | — | Generate audio (if the model supports it) |
{
"name": "generate_video",
"arguments": {
"prompt": "slow dolly-in on a neon coffee cup, steam rising, cinematic",
"aspect_ratio": "9:16",
"resolution": "720p",
"duration": 5
}
}{
"request_id": "a7b2...4ce",
"status": "IN_PROGRESS",
"model": "seedance-2-fast-t2v",
"cost_credits": 12,
"next_step": "call wait_generation with this request_id"
}Video status is awaited via wait_generation (long-poll up to 30 s per call) — the connection does not drop on long clips.
cost_credits is returned at launch — the agent knows the cost before waiting.
The agent can launch several 9:16 social clips in one chat or embed generation into headless CI.
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.
The credit cost is returned at launch (cost_credits) and depends on the model, duration, and resolution.
Video takes longer than an image. generate_video returns a request_id, and the agent fetches the result via wait_generation.
Yes — pass image_url and the tool works in image-to-video mode.
Yes, API-key setup works headless — on a server and in pipelines.