douyin-video-analyst

Douyin Video Analyst

Batch collect the latest video links from a Douyin account homepage, extract video speech copy via douyin-mcp, remove colloquialisms and output concise original copy, and provide a comprehensive summary.

Prerequisites

browser tool: used to render the Douyin SPA page and scrape the video list (web_fetch is invalid; Douyin is purely JS-rendered)
mcporter CLI：which mcporter confirmed available
douyin-mcp server: in mcporter list it should show douyin-mcp (5 tools)

mcporter will automatically load the MCP servers from the two config files ~/.cursor/mcp.json and ~/.claude.json .

Version and Key reference

Version	API platform	Environment variable
≤ 1.1.0	SiliconFlow	`DOUYIN_API_KEY`
≥ 1.2.0	Alibaba Cloud Model Studio	`DASHSCOPE_API_KEY`

Both versions are supported; choose the corresponding environment variable according to the actual configuration. For configuration guide, see references/setup.md。

Workflow steps

Step 1: Check configuration

# 确认 mcporter 和 douyin-mcp 可用
mcporter list 2>&1 | grep douyin-mcp

Read the config file to determine the version and Key:

python3 << 'EOF'
import json, os
# mcporter 同时加载这两个配置
for path in ['~/.cursor/mcp.json', '~/.claude.json']:
    p = os.path.expanduser(path)
    if not os.path.exists(p): continue
    d = json.load(open(p))
    cfg = d.get('mcpServers', {}).get('douyin-mcp')
    if cfg:
        print(f"Found in {path}")
        print('  args:', cfg.get('args'))
        print('  env keys:', list(cfg.get('env', {}).keys()))
        break
EOF

env contains DOUYIN_API_KEY → v1.1.0, SiliconFlow
env contains DASHSCOPE_API_KEY → v1.2.0+, Alibaba Cloud Model Studio

If not configured, refer to references/setup.md to complete initialization.

Step 2: Use browser to scrape the video list

Douyin is a pure JS SPA;the browser tool must be used, web_fetch cannot be used.

browser(action="open", profile="openclaw", targetUrl="<账号主页URL>")
browser(action="screenshot", ...)   # 等页面渲染，确认视频列表出现
browser(action="snapshot", ...)     # 提取 listitem 中的 /video/XXXX 链接

Extract from the snapshot's aria tree link[href=/video/XXXXXXXXXXXXXXXXX], take the first N items, and concatenate the complete links:https://www.douyin.com/video/<video_id>

Step 3: Concurrently extract video copy

According to the version detected in Step 1, choose the corresponding environment variable name,concurrentlycall all videos (do not wait serially):

# v1.1.0（硅基流动）
DOUYIN_API_KEY="<key>" mcporter call douyin-mcp.extract_douyin_text \
  share_link="https://www.douyin.com/video/<video_id>" 2>&1

# v1.2.0+（阿里云百炼）
DASHSCOPE_API_KEY="<key>" mcporter call douyin-mcp.extract_douyin_text \
  share_link="https://www.douyin.com/video/<video_id>" 2>&1

Initiate all exec calls at the same time (place them in the same tool call block), set timeout=180，yieldMs=120000。

If an error occurs, refer to references/troubleshooting.md。

Step 4: Organize and output

Output in video order (newest to oldest), each video includes:

Video title(from the snapshot's link text)
Video link
Concise copy: remove colloquial expressions from the original speech copy (remove filler words, repetitive verbosity, and colloquial filler such as "everyone"/"we"/"you"), keep the core information and arguments, present as written paragraphs, do not summarize or compress, fully retain all key points

Finally append a comprehensive summary: summarize the main logic, core themes, and consistent risk warnings across videos.

Output format reference

## 📊 [账号名] 最近 N 条视频文案整理

---

### 视频1 | <标题>
🔗 <链接>

<精简后的书面化文案，完整保留所有要点，去除口语填充>

---

### 视频2 | <标题>
...

---

## 🗂️ 综合总结

| 主线 | 核心逻辑 | 重点方向 |
|------|----------|----------|
| ... | ... | ... |

Notes

When the Douyin login popup appears: read the snapshot directly; the video list can usually still be obtained from the aria tree
The first video may require login to access; from the second one onward it is usually normal
extract_douyin_text takes a long time (30~90s per item), so be sure to execute concurrently
If copy extraction fails, you can first use parse_douyin_video_info to get the download link, then process manually