This project provides an integrated desktop toolkit for Douyin (TikTok China) content creators, automating the workflow from material acquisition to AI-driven content generation. It targets creators and marketers seeking to streamline operations, offering a local-first, privacy-conscious solution with advanced AI capabilities and a modular agent system.

How It Works

The toolkit employs a Tauri 2, React, and Rust stack to deliver a Windows desktop application. It orchestrates a pipeline involving batch local video transcription or Douyin link parsing, followed by AI analysis of video structure and content. A private knowledge base can be integrated for semantic search, enhancing AI outputs. The core innovation is the "Agent Studio," which chains these components using pre-defined "Skills" to generate structured content, scripts, or operational insights, leveraging local ASR (Sherpa-ONNX, SenseVoice) and various AI providers.

Quick Start & Requirements

Installation is available via pre-built Windows installers (.exe, .zip) or by building from source. Source builds require Windows 10/11 x64, Node.js 18+, Rust Stable, PowerShell 5.1+, Visual Studio C++ Build Tools, and Git. Initial setup involves downloading ASR and embedding models, configuring AI providers (Doubao, OpenAI, DeepSeek, LM Studio), and optionally setting up NVIDIA GPU acceleration. GitHub Releases: https://github.com/lid664951-crypto/douyin-creator-toolkit/releases

Highlighted Details

• Agent Studio: A central feature for v1.2.0, acting as a "Douyin creation workflow Agent MVP" that connects transcription, knowledge base, and AI analysis.
• 10 Built-in Agent Skills: Pre-packaged modules for tasks like script generation, hot topic ideation, text refinement ("de-AIing"), performance review, and comment response, extensible via skills.json.
• Local-First & Privacy: Core data, configurations, and tasks are stored locally. API keys are encrypted using Windows DPAPI.
• Batch Processing: Supports batch transcription of up to 50 local videos and batch extraction/download from Douyin links.
• GPU Acceleration: Integrates NVIDIA CUDA for accelerated local speech-to-text transcription.

Maintenance & Community

The provided README does not detail specific contributors, sponsorships, or community channels like Discord or Slack.

Licensing & Compatibility

The project is licensed under the MIT License, which is permissive for commercial use and integration into closed-source projects. The application is specifically designed for Windows.

Limitations & Caveats

This toolkit is primarily a Windows-only desktop application. The Agent Studio is positioned as a foundational "MVP" for specific workflows, not a general-purpose AI agent. GPU acceleration is focused on NVIDIA CUDA environments. Initial application startup may be slow due to resource initialization and model downloads.