360 Secure SkillHub
360 安全技能中心

VideoGen - AI Video Generator

该技能调用 VideoGen 托管的 MCP 工具,把文本提示、脚本、图片、录音或 PDF/演示文稿转为视频,覆盖分镜、配音、幻灯片等工作流,并可轮询任务、重混项目、导出下载链接;需用户先在宿主中完成 OAuth 连接。

videogen
热度 15.01
最近更新 2026-10-03 02:35:36
分类:创作依赖:MCP 服务 / 第三方 API应用场景:视频编辑 / 图像生成 / 故事创作 / 演示设计 ...

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add videogen-video
Metadata
name
videogen
description
Use VideoGen’s AI video generator and AI video editor for text to video, image to video, or script to video. Create, edit, caption, and export AI videos and media when the user asks to use VideoGen.

VideoGen

Use the hosted VideoGen tools to produce media and downloadable videos.

If the host has no VideoGen MCP connection, configure the remote HTTP endpoint https://mcp.videogen.io/mcp in its MCP settings and complete OAuth sign-in. Installing this standalone skill does not install the connector. Setup and service details: https://videogen.io/videogen-mcp.

Connect and discover

Call get_me to check the connection. If authentication is required, ask the user to connect their VideoGen account through the host's OAuth flow. Do not ask them to paste credentials into chat.

Call get_getting_started_guidance for onboarding, get_workflows_guidance before a full video workflow, and get_async_tasks_guidance before polling. Use the current tool schemas and guidance for accepted inputs and actions. Full video generation commonly takes several minutes and uses account credits; respect the user's requested scope and budget.

Choose a workflow

Intent Tool
Product ad or shot-by-shot video storyboard_to_video
Narrated video from a script script_to_video
Video around recorded narration voiceover_to_video
Video from a PDF or presentation slideshow_to_video
One generated clip prompt_to_video_clip
Individual image, voice, music, or sound effect Matching generate_* tool

For a product ad, a useful default is vertical 9:16 with a hook, product in use, and call to action. Preserve the user's chosen format and scene count. Storyboard visuals belong in each scene's prompt; spoken copy belongs in voiceoverScript. Prefer a full workflow to manually chaining individual media tools when the user wants a complete video.

Transfer files

open_uploader is a ChatGPT-only interface and is unavailable in these agent hosts. Use upload_file for small files when the host can supply base64 bytes. For larger files, use create_file_upload, transfer the bytes to the returned URL when supported, then call get_file with wait: true. If the host cannot transfer an attachment, direct the user to upload it in the VideoGen app. Never claim upload success without a returned VideoGen fileId.

Wait, edit, and export

Poll get_workflow_run for workflows, get_tool_execution for media, list_project_remix_actions for remixes, and get_project_export for exports. Follow the async guidance until a terminal status; never invent completed results, ids, or URLs.

Preserve the returned projectId. For requested edits, use remix_project with actions supported by its current schema, then wait for completion before calling export_project. Use saveAsNewProject: true if the user wants to preserve the original. Return the successful export's downloadUrl. If an edit is unavailable, use get_app_deep_link to open the project in VideoGen.

Report terminal failures clearly and retry only as directed by structured errors or guidance. For credit or account-access errors, direct the user to https://app.videogen.io. Do not expose internal model or infrastructure names.