← 返回 Skills 市场
126
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install douyin-video-to-report
功能描述
抖音视频链接 → 完整图文报告。自动完成:绕过验证抓取视频 → 下载 → 音频提取 → 语音转文字 → 炫酷HTML报告 → 飞书发送。触发词:抖音视频总结、抖音视频分析、这个视频说了什么、总结这个视频
使用说明 (SKILL.md)
Douyin Video → Report Pipeline
将抖音视频链接自动转换为可阅读的 HTML 报告,适合需要提炼视频内容的场景。
完整工作流程
抖音短链 (v.douyin.com/xxx)
│
▼
1. Playwright (headless Chrome)
- 绕过 JS 验证 / 滑动验证码
- 提取视频直链
│
▼
2. curl 下载视频 (.mp4)
│
▼
3. ffmpeg 提取音频 (.mp3)
│
▼
4. miaoda-studio-cli speech-to-text
- 中文普通话识别
- 100% 转录
│
▼
5. 生成炫酷 HTML 报告
- 深色渐变主题
- 核心数据卡片
- 洞察总结
- 完整转录(可展开)
- Ctrl+P 可打印成 PDF
│
▼
6. 飞书附件发送给用户
核心脚本
脚本位于:/home/gem/workspace/agent/skills/douyin-video-to-report/douyin_pipeline.py
使用方式
# 方式一:直接运行(传入抖音URL)
python3 /home/gem/workspace/agent/skills/douyin-video-to-report/douyin_pipeline.py "https://v.douyin.com/xxxxx"
# 方式二:分步执行(见下方分步说明)
分步执行命令
Step 1:提取视频直链
# douyin_fetch.py - Playwright 提取视频 URL
import asyncio
from playwright.async_api import async_playwright
async def get_video_url(douyin_url):
async with async_playwright() as p:
browser = await p.chromium.launch(headless=True, args=[
'--no-sandbox', '--disable-setuid-sandbox',
'--disable-blink-features=AutomationControlled',
])
context = await browser.new_context(
user_agent='Mozilla/5.0 (iPhone; CPU iPhone OS 17_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.0 Mobile/15E148 Safari/604.1',
viewport={'width': 390, 'height': 844},
locale='zh-CN',
)
page = await context.new_page()
await page.goto(douyin_url, timeout=30000, wait_until="networkidle")
await asyncio.sleep(3)
# 从 video 元素获取直链
video_info = await page.evaluate("""
() => {
const v = document.querySelector('video');
if (v && v.src) return v.src;
return null;
}
""")
await browser.close()
return video_info
Step 2:下载视频
curl -L -o /tmp/douyin_video.mp4 "视频直链" \
-H "User-Agent: Mozilla/5.0 (iPhone; CPU iPhone OS 17_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.0 Mobile/15E148 Safari/604.1" \
--max-time 60 -s
Step 3:提取音频
ffmpeg -i /tmp/douyin_video.mp4 -vn -acodec libmp3lame -q:a 2 /tmp/douyin_audio.mp3 -y
Step 4:语音转文字
miaoda-studio-cli speech-to-text --file /tmp/douyin_audio.mp3 --lang zh --output text
Step 5:生成报告
使用 miaoda_coding 或直接使用 report_generator.py(见下方)
依赖环境
| 依赖 | 安装命令 |
|---|---|
| Playwright | pip install playwright && python3 -m playwright install chromium |
| ffmpeg | 系统自带或 apt install ffmpeg |
| yt-dlp | pip install yt-dlp |
| miaoda-studio-cli | 系统自带 |
关键经验
- 必须用 Playwright headless Chrome,不能用 requests/httpx,抖音会拦截
- 移动端 UA 更容易绕过,用 iPhone UA 而非桌面 Chrome
- 视频直链从
document.querySelector('video').src获取,不要从 network 请求里找 - 下载视频时必须加 User-Agent header,否则 403
- 飞书附件发送:文件必须放在
/home/gem/workspace/agent/workspace/或/tmp/openclaw/目录
报告模板结构
- Hero:标题 + 视频元信息
- 数据卡片:关键数字高亮
- 洞察区:6个核心发现卡片
- 对比表:进门 vs 留存因素
- 案例区:引用高亮
- 行动指南:5项建议
- 完整转录:可展开/收起
相关工具
miaoda-studio-cli speech-to-text:音频转文字miaoda-studio-cli doc-parse:文档解析miaoda-coding:生成可视化报告
安全使用建议
This skill largely does what it claims (scrape Douyin → extract audio → transcribe → analyze frames), but it has several red flags you should consider before installing or running it: 1) The README promises HTML report generation and automatic Feishu sending, but the included script does not implement those steps — ask the author for the missing implementation or remove any claims you don't want. 2) The script calls unknown external CLIs (miaoda-studio-cli / miaoda-coding); these could perform network calls or require credentials that the skill doesn't declare — verify their provenance and what data they send. 3) The code uses subprocess.run(..., shell=True) with interpolated inputs (URLs/paths) — this is a command-injection risk if inputs are attacker-controlled; run only with trusted inputs or after sanitizing. 4) The script uses Playwright with anti-detection flags to bypass site protections — be mindful of legal/ethical implications of scraping and of running such code in production. Recommended actions: run the script in an isolated sandboxed environment with restricted network egress, audit or replace the miaoda-* tooling, ensure inputs are validated, and ask the publisher to reconcile the documented features with the actual code (especially Feishu sending and report generation).
功能分析
Type: OpenClaw Skill
Name: douyin-video-to-report
Version: 1.0.0
The skill bundle implements a Douyin video scraping and analysis pipeline but contains significant security flaws. Specifically, `douyin_pipeline.py` uses `subprocess.run(shell=True)` with unsanitized input (the video URL), creating a high risk of command injection (RCE). While the script's behavior aligns with its stated purpose of video processing and report generation, its use of Playwright to bypass bot detection and the lack of input validation on shell commands make it a security risk in an agentic environment.
能力评估
Purpose & Capability
Name/description match the included code: Playwright capture, ffmpeg extraction, frame analysis and transcription. However the SKILL.md and description promise extra capabilities (automatic HTML report generation, 飞书 attachment sending) that are not implemented in the provided douyin_pipeline.py (script stops after saving transcript/analysis). The README also lists tools (yt-dlp, miaoda-coding) that the code does not call. These inconsistencies mean the package does not fully implement or document its claimed end-to-end behavior.
Instruction Scope
Runtime instructions and code perform web scraping with Playwright (including anti-detection flags to '绕过 JS 验证'), download remote video URLs, write files under /tmp and /home/gem/workspace/agent/workspace, call external CLIs (miaoda-studio-cli) for transcription and image analysis. The SKILL.md explicitly instructs storing files in specific workspace paths and indicates automatic sending to 飞书 (not implemented). The instructions give broad discretion for bypassing site protections which is scope-creep relative to a simple 'summarize' task and could be abused.
Install Mechanism
There is no install spec (instruction-only skill plus a script). That minimizes supply-chain risk from package downloads. The SKILL.md lists dependencies and install commands for Playwright/ffmpeg/yt-dlp, which are reasonable for the task.
Credentials
The skill declares no required environment variables or credentials, but invokes external tooling (miaoda-studio-cli, miaoda-coding) of unknown provenance which likely perform network I/O and may require credentials or leak data; these are not declared. The code writes into a hard-coded workspace path (/home/gem/...), which may be harmless in a dedicated agent environment but could expose data if run on a shared host. Also the script uses subprocess.run(shell=True) with f-strings that include remote-derived data (video URLs), creating a command-injection risk if inputs are maliciously crafted.
Persistence & Privilege
The skill is not always-enabled and does not modify other skills or global agent configuration. It runs on invocation only. No explicit persistence or elevated privileges are requested in the manifest or code.
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-to-report - 安装完成后,直接呼叫该 Skill 的名称或使用
/douyin-video-to-report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 douyin-video-to-report: convert Douyin video links into comprehensive HTML reports.
- Automated pipeline: bypass verification, download video, extract audio, speech-to-text, generate stylish report, and send via Feishu.
- Includes detailed step-by-step commands and dependency setup.
- Provides robust tips for handling Douyin’s technical restrictions and best practices for accurate video processing.
- Customizable report template features key data highlights, insights, full transcript, and actionable advice.
元数据
常见问题
抖音视频转报告 是什么?
抖音视频链接 → 完整图文报告。自动完成:绕过验证抓取视频 → 下载 → 音频提取 → 语音转文字 → 炫酷HTML报告 → 飞书发送。触发词:抖音视频总结、抖音视频分析、这个视频说了什么、总结这个视频. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 126 次。
如何安装 抖音视频转报告?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install douyin-video-to-report」即可一键安装,无需额外配置。
抖音视频转报告 是免费的吗?
是的,抖音视频转报告 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
抖音视频转报告 支持哪些平台?
抖音视频转报告 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 抖音视频转报告?
由 wahtungL(@wahtungl)开发并维护,当前版本 v1.0.0。
推荐 Skills