SceneLoop

English | 中文文档（macOS） | 中文文档（Windows）

Install

Recommended: One-Command Install with npx

We recommend running SceneLoop directly with npx; no global installation is required:

npx --yes sceneloop

The unified installer detects the operating system and architecture, downloads only the matching platform package, detects installed agents, and asks where to install SceneLoop.

Alternative: Install Globally with npm

Global npm installation is also supported:

npm install -g sceneloop
sceneloop

Installation Options

Install into one, several, or all detected agents:

npx --yes sceneloop --agent hermes
npx --yes sceneloop --agent claude,codex
npx --yes sceneloop --all --yes

Inspect detection or preview changes without writing files:

npx --yes sceneloop --list
npx --yes sceneloop --all --dry-run

The original platform-specific commands remain available:

npx --yes sceneloop-mac
npx --yes sceneloop-win

When updating an existing installation, SceneLoop preserves local configuration and user project data and backs up the previous packaged files.

SceneLoop is a proprietary, binary-distributed AI production Skill for creating AI dramas, single-image animations, and AI product ads. It works with Hermes, OpenClaw, Claude Code, and Codex, while SceneLoop's configured production models perform the actual text, image, and video generation.

Developed by Xi'an Wenyao Network Information Technology Co., Ltd.

SceneLoop Product Website

What SceneLoop Can Do

• Turn DOCX, Markdown, or TXT scripts into structured AI drama projects.
• Plan storyboards, characters, locations, shots, and visual continuity.
• Generate character references, location references, first frames, and videos.
• Render individual shots or complete episodes in storyboard order.
• Resume incomplete projects and retry only missing or failed work.
• Animate one uploaded image directly from a motion or camera description.
• Preserve completed assets instead of regenerating successful work.
• Create one grounded 15-second Fast UGC product ad from one to five product images or commerce-page screenshots.
• Resume interrupted AI Ads projects without resubmitting successful provider tasks or replacing completed assets.

Workflow

Script
  → Visual adaptation
  → Storyboard, characters, and locations
  → Character and location references
  → Shot first frames
  → Shot videos
  → Completed episode

Single-image animation uses a shorter workflow:

Uploaded image + motion description
  → Image animation
  → Video output

AI product advertising uses a separate workflow:

Product images or commerce screenshot + short brief
  → Grounded product analysis and creative planning
  → Five image assets
  → Five three-second video clips
  → One 15-second Fast UGC preview

Drama and image animation run through the packaged sceneloop binary. Product advertising runs through the packaged ai-ads binary. Both share the packaged setup program, local licensing, and model configuration, while keeping their project data separate.

Supported Platforms

• macOS 14 or later on Apple Silicon (arm64)
• Windows 10 or 11 (x64)

Intel Mac, Windows ARM64, and Linux are not currently supported.

Requirements

• Node.js 18 or later and npm 9 or later
• Hermes, OpenClaw, Claude Code, or Codex
• A valid SceneLoop License Key
• API keys for the selected Vision, Text, Image, and Video providers
• ffmpeg and ffprobe available in PATH when assembling AI Ads videos

Python is not required when using the packaged executables.

Each platform package installs three native programs:

| Purpose | macOS | Windows | |---|---|---| | Drama and image animation | scripts/sceneloop | scripts/sceneloop.exe | | AI product advertising | scripts/ai-ads | scripts/ai-ads.exe | | License and model setup | scripts/setup | scripts/setup.exe |

License Activation

SceneLoop requires a valid License Key. To purchase or request one, contact the SceneLoop team by email or WeChat:

• Email: [email protected]
• Product website: SceneLoop

| WeChat: 1.382 | WeChat: 漂泊西贝 | | :---: | :---: | | | |

The License Key is not included in the npm package. After receiving it, enter it only in the local SceneLoop Setup flow and do not commit it to source control.

Start Using SceneLoop

After installation, start a new agent session and ask the agent to use SceneLoop:

Use SceneLoop to turn this script into the first episode of an AI drama.

For one-image animation:

Use SceneLoop to animate this image. Make the subject blink gently, add slight
hair movement, and slowly push the camera forward.

For a product ad:

Use SceneLoop to create a natural 15-second vertical UGC ad from this product
screenshot. Generate it directly.

The current Fast UGC workflow uses one to five product images, creates five three-second shots, and assembles one 15-second preview. One clear commerce screenshot is sufficient; users do not need to transcribe its visible product name, price, or specifications. Before paid generation, the agent confirms the complete scope once. Interrupted work resumes from the same project.

SceneLoop checks local licensing and provider configuration before production. If setup is incomplete, it launches the packaged setup program. License Keys and API keys are entered only in that local interactive setup—not in agent chat.

Security

• Enter License Keys and API keys only in the local SceneLoop setup prompt.
• Never send credentials through an agent conversation, group chat, or log.
• Do not modify or bypass SceneLoop license validation.
• The npm packages do not contain user credentials.

Maintainer Commands

npm test
npm run pack:check

npm publish --workspace packages/macos --registry=https://registry.npmjs.org/
npm publish --workspace packages/windows --registry=https://registry.npmjs.org/
npm publish --workspace packages/cli --registry=https://registry.npmjs.org/

Publish both platform packages before publishing the unified sceneloop entry package.

License

SceneLoop is distributed under a Proprietary License. The complete license is included in the npm package as the LICENSE file.

Copyright © 2024–2026 Xi'an Wenyao Network Information Technology Co., Ltd.