Video analyze by doubao2.0

What this skill does

Privacy and data-safety notice: the skill processes video via the Volcengine Ark API (https://ark.cn-beijing.volces.com) — local video files are base64-encoded and uploaded to Volcengine servers, and remote video URLs are sent directly to Volcengine for parsing; ensure the video contains no sensitive, confidential, or unauthorized content, and data handling follows Volcengine's terms of service and privacy policy. Prerequisites: verify the ARK_API_KEY environment variable exists (terminate and guide configuration if missing), then run pip install 'volcengine-python-sdk[ark]' to ensure dependencies. Workflow: first clarify the analysis goal and write a high-quality prompt yourself (the "role/task/output" three-part template is a reference, not mandatory); then call the script — local video python script/parse_video.py --video 'absolute/video/path.mp4' --prompt 'your prompt content', remote URLs add --remote; wait for the command to finish and take the result. Note: if a remote network video fails to execute, notify the user of the error content directly and NEVER retry with web_fetch-like tools. When a network video exceeds the size limit, first download it into the workspace's video directory, then invoke it as a local video. The skill card warns that selected local videos or remote URLs are sent to Volcengine Ark for processing (use only for videos approved for third-party processing and avoid sensitive, confidential, or unauthorized content), that the script requires ARK_API_KEY and exposes the credential to the active runtime (run in an isolated environment, scope the key appropriately, avoid logging or sharing it), that setup installs an unpinned SDK into the active Python environment (use a virtual environment and pin a reviewed volcengine-python-sdk[ark] version before deployment), and that remote videos may be downloaded into the workspace when direct remote parsing fails (delete downloaded files after use and apply normal workspace data-handling controls).

Profile published 2026-10-27

Example prompts

Generated from the skill’s documentation as examples; not results of actual runs.

Monthly Trend

SKILL.md (original)

The author’s original file, shown as plain text without edits or review by AlphaMoat. Links are not clickable; check commands before running them. Fetched 2026-10-23.

---
name: doubao-video-analyze
description: 使用豆包2.0模型解析视频。当需要执行分析视频内容等需要理解视频视觉信息时调用该技能。你必须在持有本地视频路径或网络视频链接时才能调用该技能
metadata: {"clawdbot":{"emoji":"🔍","requires":{"bins":["python"],"env":["ARK_API_KEY"]},"primaryEnv":"ARK_API_KEY"}}
---

# Doubao Video Analyze

## 隐私与数据安全声明

**重要**：本技能通过火山引擎 Ark API（https://ark.cn-beijing.volces.com）处理视频内容。使用此技能时：

- 本地视频文件将被编码为 base64 并上传至火山引擎服务器
- 远程视频 URL 将直接发送给火山引擎服务器进行解析
- 请确保视频内容不包含敏感、机密或未经授权的信息
- 视频数据的处理遵循火山引擎的服务条款和隐私政策

## 使用方法

### 注意

对于远程网络视频，如果通过该技能无法执行，请直接通知用户错误内容，**永远不要**尝试使用web_fetch之类的工具去重试！

### 前置检查

1. 检查当前是否有`ARK_API_KEY`环境变量，如果没有，直接终止并引导用户先进行配置
2. 直接尝试调用如下bash指令，确保用户当前python环境有openai依赖

```bash
pip install 'volcengine-python-sdk[ark]'
```

### 分析视频

1. 你需要先明确你的分析目的是什么，你需要自行写出一段提示词，例如

```plaintext
### 角色
你是一位xxxxx，善于分析视频的xxxxxx

### 任务
你的任务是xxxxxxx

### 输出
你需要输出以下内容：
xxxxxxxxxxxxxx
```

**注意**：你需要写出高质量的提示词，但并不一定要按照上面的模板

2. 调用本技能目录下脚本以获取分析结果

- 本地视频

  ```bash
  python script/parse_video.py --video 'absolute/video/path.mp4' --prompt 'your prompt content'
  ```

  

- 远程视频链接

  ```bash
  python script/parse_video.py --video 'https://example.com/video_url.mp4' --prompt 'your prompt content' --remote
  ```




等待该命令执行结束并获取结果作为该视频的分析结果



### 网络视频错误处理

如果遇到网络视频超过大小限制的问题，首先现将该网络视频下载至你的workspace的video目录下，然后再通过本地视频的方式调用

Popular in Design & Media