视频编辑器

• Gemini Omni Flash
• Grok Imagine Video

示例

前

后

前 / 后Grok Imagine Video

夜晚 → 金色黎明（电影感）

Grok Imagine Video。超宽画幅的夜间滑翔被重新想象成云海之上的金色日出：一条指令完成整体环境与光线的改写，而 Pyra 依旧清晰、未被改动。

复现打开

视频编辑器 API

从你自己的代码调用此流水线。一次请求即可发起运行；模型是一个输入字段，而不是独立的接口。

import { createClient } from "@pipe2-ai/sdk";

const client = createClient(process.env.PIPE2_TOKEN!);

const { run_pipeline } = await client.RunPipeline({
  pipeline_slug: "video-editor",
  input: {
    video_url: "https://example.com/interview.mp4",
    instructions: "Cut the dead air at the start and add a slow zoom on the speaker",
    model: "gemini-omni-flash",
  },
});

import asyncio
import os

from pipe2 import Pipe2Client

client = Pipe2Client(token=os.environ["PIPE2_TOKEN"])

run = asyncio.run(client.run_pipeline(
    pipeline_slug="video-editor",
    input={
        "video_url": "https://example.com/interview.mp4",
        "instructions": "Cut the dead air at the start and add a slow zoom on the speaker",
        "model": "gemini-omni-flash",
    },
))

client := pipe2.NewClient(os.Getenv("PIPE2_TOKEN"))

input := json.RawMessage(`{
  "video_url": "https://example.com/interview.mp4",
  "instructions": "Cut the dead air at the start and add a slow zoom on the speaker",
  "model": "gemini-omni-flash"
}`)

run, err := pipe2.RunPipeline(context.Background(), client, "video-editor", input)
if err != nil {
    log.Fatal(err)
}

pipe2 pipelines run \
  --pipeline video-editor \
  --input-json '{"video_url": "https://example.com/interview.mp4", "instructions": "Cut the dead air at the start and add a slow zoom on the speaker", "model": "gemini-omni-flash"}' \
  --wait

获取 API 令牌 →

参数

instructions 必填

What to change, in plain language.

string

video_url 必填

Public URL of the video to edit.

string

model

Video engine to use. Leave unset and Pipe2 picks one that supports your edit.

gemini-omni-flashgrok-imagine-video 默认 gemini-omni-flash

在 AI 智能体中使用

将任意 MCP 客户端指向 pipe2，你的智能体就能获得这些工具。它会找到这条流水线，读取上方相同的 schema，然后运行它。

• list_pipelines
• get_pipeline_schema
• run_pipeline
• get_pipeline_run_status
• request_upload

{
  "mcpServers": {
    "pipe2ai": {
      "url": "https://mcp.pipe2.ai/mcp",
      "headers": { "Authorization": "Bearer YOUR_TOKEN" }
    }
  }
}

完整 API 参考→ MCP 设置→ CLI 参考→

常见问题