← 返回 Skills 市场
67
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install wansan-video-summarizer
功能描述
Multi-platform video transcript extraction and AI-powered summarization (YouTube, Bilibili, extensible). Use when you need to summarize videos, extract trans...
安全使用建议
This skill appears to implement what it claims, but review and decide before installing: 1) setup.sh will create a venv and pip-install packages—inspect and run it yourself rather than blindly piping to shell. 2) yt-dlp is invoked with --cookies-from-browser chrome: that can cause local browser cookies (sensitive) to be read by yt-dlp; only allow this if you understand and accept that access. 3) The SKILL.md mentions using an 'innertube + Cloudflare proxy' technique to avoid rate limits—ask the author which endpoints/proxies are used or inspect code paths that call innertube to ensure no unexpected network endpoints are contacted. 4) Only provide optional tokens (LLM_API_KEY, GITHUB_TOKEN, OPENCLAW_GATEWAY_TOKEN, POLLINATIONS_API_KEY) if you trust the skill and intend to use those backends. If you want higher assurance, request the full summarize.py logic that performs LLM and network calls (the included file was truncated) and check for any hard-coded remote endpoints or obfuscated network behavior.
功能分析
Type: OpenClaw Skill
Name: wansan-video-summarizer
Version: 2.0.0
The skill exhibits high-risk behaviors including accessing local browser data and utilizing sensitive authentication tokens. Specifically, 'scripts/summarize.py' invokes 'yt-dlp' with the '--cookies-from-browser chrome' flag to bypass Bilibili's anti-bot measures, which grants the script access to the user's browser history and session data. Additionally, the script contains logic to utilize a 'GITHUB_TOKEN' by spoofing VS Code/Copilot headers to access the GitHub Copilot API as an LLM fallback. While these actions appear aligned with the stated purpose of video summarization and bypassing platform restrictions, the combination of browser data access and sensitive token usage represents a significant security risk without explicit user consent for these specific high-privilege operations.
能力标签
能力评估
Purpose & Capability
Name/description align with included code and dependencies (yt-dlp, ffmpeg, faster-whisper, youtube-transcript-api, innertube). Using yt-dlp + faster-whisper for Bilibili transcription and ffmpeg for frames is coherent. One minor mismatch: the README/SKILL.md emphasize an 'innertube ANDROID client + Cloudflare proxy' technique to avoid YouTube limits but the code excerpt doesn't show any explicit proxy endpoints or configuration—this is ambiguous but could be an implementation detail of the innertube library.
Instruction Scope
Runtime instructions and scripts will: (a) invoke yt-dlp with --cookies-from-browser chrome (this allows access to local browser cookies), (b) run ffmpeg and external subprocesses, (c) create files under /tmp and a skill-local config/settings.json, and (d) (per README) may call external LLM/image endpoints (LLM_API_URL, OPENCLAW_GATEWAY_TOKEN, GITHUB_TOKEN, POLLINATIONS_API_KEY) if provided. Access to browser cookies and unspecified proxy behavior is the primary scope creep risk; these are not required env vars but are implied by the download approach.
Install Mechanism
There is no registry install spec, but a bundled setup.sh installs a Python venv and pip packages (youtube-transcript-api, requests, innertube, faster-whisper) and checks for yt-dlp/ffmpeg. This is a common pattern; it uses pip (PyPI) and Homebrew for yt-dlp if needed—moderate risk but expected. Nothing in setup.sh downloads arbitrary archives from untrusted servers.
Credentials
Requires no credentials to run, but documents many optional environment variables (LLM_API_URL/KEY, OPENCLAW_GATEWAY_TOKEN, GITHUB_TOKEN, POLLINATIONS_API_KEY). Those are plausible for optional LLM/image fallbacks, but supplying broad tokens like GITHUB_TOKEN or gateway tokens increases blast radius. The SKILL.md/README do not require these; they are optional and explained as fallbacks.
Persistence & Privilege
The skill does not request always:true and is user-invocable. setup.sh writes only to its own skill config directory. It does not appear to modify other skills or global agent settings.
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 wansan-video-summarizer - 安装完成后，直接呼叫该 Skill 的名称或使用
/wansan-video-summarizer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
v2.0.0
Renamed from youtube-summarizer; cross-platform install; YouTube + Bilibili support
元数据
常见问题
Video Summarizer 是什么？
Multi-platform video transcript extraction and AI-powered summarization (YouTube, Bilibili, extensible). Use when you need to summarize videos, extract trans... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 67 次。
如何安装 Video Summarizer？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install wansan-video-summarizer」即可一键安装，无需额外配置。
Video Summarizer 是免费的吗？
是的，Video Summarizer 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Video Summarizer 支持哪些平台？
Video Summarizer 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 Video Summarizer？
由 mcdowelll（@mcdowell8023）开发并维护，当前版本 v2.0.0。
推荐 Skills