← 返回 Skills 市场

Douyin Video Transcriber

Name: Douyin Video Transcriber
Author: ahsbnb

作者 an · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ⚠ suspicious

146

总下载

当前安装

版本数

在 OpenClaw 中安装

/install douyin-video-transcriber

功能描述

(已验证) 强大的抖音视频批量转写器，集成了下载、音频提取和本地 Whisper 模型转写功能。

使用说明 (SKILL.md)

SKILL.md for video-transcriber

Description

这是一个功能强大且经过实战检验的抖音视频批量转写技能。它将视频下载、音频提取和本地 AI 模型 (OpenAI Whisper) 转写三个核心步骤整合进一个自动化的工作流中，能够高效地将一个或多个抖音视频的语音内容，提取为纯文本。

该技能已被完全重构，实现了良好的可移植性和安全性，是进行视频内容分析、文案提取等任务的基础工具。

⚠️ Prerequisites (前置条件)

在使用本技能前，您必须在您的系统上完成以下配置：

安装 ffmpeg: 本技能需要使用 ffmpeg 工具来从视频中提取音频。
- 请从 ffmpeg 官网下载适合您操作系统的版本。
- 安装后，必须将 ffmpeg 的可执行文件路径添加到您系统的 PATH 环境变量中，以确保本技能可以从任何位置调用它。
配置 TikHub Token: 视频下载功能需要一个有效的 TikHub API Token。请在您的 ~/.openclaw/config.json 文件中添加 tikhub_api_token。
```
{
  "tikhub_api_token": "YOUR_TIKHUB_API_TOKEN"
}
```

How to Use

该技能的核心脚本 (scripts/run.py) 是一个功能完善的命令行工具。

Parameters

--url (必填, 可重复): 要转写的抖音视频的分享链接。您可以多次使用此参数以进行批量处理。
--output-file: 指定输出报告的 Markdown 文件路径。如果省略，将自动在 workspace/data/video-transcriber/ 目录下生成一个带时间戳的报告文件。

Example Invocation

转写单个视频:

# AI应动态查找 python 路径
python path/to/run.py --url "https://v.douyin.com/iY8K3Nqk/" --output-file "report.md"

批量转写多个视频:

# AI应动态查找 python 路径
python path/to/run.py --url "URL_1" --url "URL_2" --url "URL_3"

在批量处理时，建议省略 --output-file 参数，让技能自动生成报告文件。

安全使用建议

This skill appears to implement the advertised workflow, but take these precautions before installing: - The script requires a TikHub API token (tikhub_api_token). The registry metadata did not declare this — you must either add tikhub_api_token to ~/.openclaw/config.json or set TIKHUB_API_TOKEN in your environment. Verify the token's origin and permissions before storing it. - Review the contents of any ~/.openclaw/config.json the skill will read. The script searches for an OpenClaw root (up to 5 parent directories from the skill location, then ~/.openclaw) and will read that config file; do not keep unrelated secrets there. - ffmpeg and the Whisper Python package are required and run locally; Whisper models can be resource- and disk-intensive — ensure your machine has sufficient resources. - The skill's listed source/homepage are effectively placeholders and the publisher is unknown; if you need stronger assurance, request the original upstream repository or a signed release and manually review scripts/run.py (which you have here). - If you are uncomfortable placing a third-party token in your OpenClaw config, consider running this in an isolated environment/container or creating a dedicated OpenClaw config with only the TikHub token. Given the metadata omissions and unknown origin, proceed only after verifying the token, config file contents, and author/source.

功能分析

Type: OpenClaw Skill Name: douyin-video-transcriber Version: 1.0.0 The skill bundle is a legitimate tool for downloading and transcribing Douyin videos using the TikHub API and local OpenAI Whisper models. The Python script `scripts/run.py` securely handles configuration by reading a specific API token from the standard OpenClaw config path and uses `subprocess.run` safely to invoke `ffmpeg` without shell execution risks.

能力评估

ℹ Purpose & Capability

Name/description, SKILL.md and scripts/run.py are coherent: the code downloads Douyin videos from TikHub, extracts audio with ffmpeg, and transcribes locally using openai-whisper. However the registry metadata claims no required config paths or env vars while the code requires a tikhub_api_token (in ~/.openclaw/config.json or env TIKHUB_API_TOKEN). The skill also expects an .openclaw root (searching up to 5 parent dirs then falling back to ~/.openclaw) which was not declared in the metadata.

ℹ Instruction Scope

SKILL.md and the script scope are narrowly focused on downloading, extracting, and transcribing. The instructions explicitly ask the user to put tikhub_api_token into ~/.openclaw/config.json; the code reads that config file and writes output into OPENCLAW_ROOT/workspace/data/video-transcriber. The only external network calls are to the TikHub API and to video play URLs. Minor concern: the code searches parent directories for a directory containing config.json and a skills directory — it will read an OpenClaw config.json it finds, so users should ensure that file doesn't contain unrelated secrets.

✓ Install Mechanism

No install spec (instruction-only with an included script). Dependencies are typical (python, requests, openai-whisper, ffmpeg). Nothing is downloaded or executed from arbitrary URLs by an installer step.

⚠ Credentials

The runtime requires a TikHub API token (tikhub_api_token) provided either in ~/.openclaw/config.json or via TIKHUB_API_TOKEN, but the registry metadata lists no required env vars or config paths — this mismatch is a packaging/information risk. Aside from that single token, the script does not request additional credentials. Users should verify the token's scope and avoid storing unrelated secrets in the same config file.

✓ Persistence & Privilege

The skill is not always-enabled and does not alter other skills or system-wide settings. It writes reports to OPENCLAW_ROOT/workspace/data/video-transcriber and temporary files in the system temp directory (which it cleans up). It runs locally and uses local Whisper models; no privileged system changes observed.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install douyin-video-transcriber
安装完成后，直接呼叫该 Skill 的名称或使用 /douyin-video-transcriber 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release of video-transcriber – a robust batch transcription tool for Douyin videos. - Integrates video downloading, audio extraction (via ffmpeg), and local transcription using Whisper. - Automates the workflow for converting multiple Douyin video links into text transcripts. - Requires ffmpeg and TikHub API Token setup for full functionality. - Outputs transcription reports as Markdown files, with support for batch processing. - Command-line interface with flexible parameter options.

元数据

Slug douyin-video-transcriber

版本 1.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题