← 返回 Skills 市场
Yt2bb
作者
Agents365.ai
· GitHub ↗
· v2.5.0
· MIT-0
181
总下载
1
收藏
0
当前安装
7
版本数
在 OpenClaw 中安装
/install yt2bb
功能描述
Use when the user wants to repurpose a YouTube video for Bilibili, add bilingual (English-Chinese) subtitles to a video, or create hardcoded subtitle version...
安全使用建议
This skill is internally consistent with its stated purpose, but review these practical considerations before installing:
- It uses yt-dlp with --cookies-from-browser chrome: running the download step may read your Chrome cookie store to access private/age-restricted videos. If you don't want the skill to access browser cookies, avoid that flag or supply an explicit cookie file.
- The translation/publish metadata step is performed by the agent/AI — that may invoke a remote model/service depending on your agent configuration, so be mindful of what text (transcripts, video content) you allow the agent to send externally.
- The repo contains a large Python CLI (srt_utils.py) and a shell preview script; both are local and readable—inspect them if you want to verify behavior. There are no hidden network sinks or arbitrary remote downloads in the install spec.
- yt-dlp, whisper (openai-whisper), and ffmpeg are powerful tools: only use binaries from trusted sources and keep them up to date.
- Consider copyright/legal implications of downloading and republishing YouTube content to other platforms.
If you want stronger containment: run the skill in an environment without Chrome login or remove the --cookies-from-browser usage, and ensure your agent's model invocation settings restrict remote calls or logging of sensitive transcripts.
功能分析
Type: OpenClaw Skill
Name: yt2bb
Version: 2.5.0
The skill bundle contains instructions in SKILL.md to execute high-risk commands, specifically 'yt-dlp --cookies-from-browser chrome', which accesses sensitive browser data. It also includes a 'Pre-flight' update check that performs a broad search across multiple hidden directories in the user's home folder and initiates network activity via 'git fetch'. While these actions are plausibly aligned with the stated purpose of downloading restricted videos and skill maintenance, they represent significant security risks and potential vectors for data exfiltration or reconnaissance if misused.
能力标签
能力评估
Purpose & Capability
Name/description (YouTube→Bilibili, EN↔ZH subtitles) lines up with required binaries and included code: yt-dlp (download), whisper (transcribe), srt_utils.py (subtitle processing), ffmpeg (burn-in). Required tools and scripts are appropriate and expected for the declared workflow.
Instruction Scope
SKILL.md keeps to the stated pipeline (download → transcribe → translate → merge → burn → generate metadata). It also: (a) searches common agent skill directories to find SKILL_DIR and runs a git fetch to check for updates (it does not auto-pull), and (b) uses yt-dlp with --cookies-from-browser chrome which reads browser cookie stores to enable downloading content requiring login. Both behaviors are coherent with the purpose but worth noting because they touch local browser data and the filesystem locations of other agent skills.
Install Mechanism
Install entries are standard package installs (brew/apt for ffmpeg, pip for yt-dlp and openai-whisper). No arbitrary URL downloads, no extract-from-unknown-host steps. The shipped Python script (srt_utils.py) is included in the repo and visible for review.
Credentials
The skill declares no required environment variables or external credentials (correct). However, yt-dlp usage documented relies on extracting Chrome cookies from the local browser profile to access age-restricted/private videos — this is functionally necessary for some downloads but does imply access to local browser auth data. Also translation/publish metadata steps are left to the agent (AI), which may be a remote model call depending on how your agent is configured.
Persistence & Privilege
always:false and user-invocable:true (normal). The only cross-repository/system actions are a read/search for SKILL_DIR and a conditional git fetch; the skill does not auto-pull or modify other skills or system-wide configs.
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 yt2bb - 安装完成后，直接呼叫该 Skill 的名称或使用
/yt2bb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
v2.5.0
Netflix-grade bilingual subtitles: new netflix ASS preset, lint subcommand enforcing Netflix Timed Text Style Guide (CPS/duration/line-length/gap checks), tightened SC translation rules, resolution-aware scaling, README preset previews. Fixes latent ASS Style field-ordering bug and clean preset colors. Breaking: cinema preset removed (use netflix or clean). Full notes: https://github.com/Agents365-ai/yt2bb/releases/tag/v2.5.0
v2.4.0
Multi-pass fix_srt for cascading overlaps. --dry-run for merge and to_ass. metadata.json as editable source for SKILL.md frontmatter.
v2.3.2
Add cross-platform install entries (apt, pip) for Linux/Windows. Add Privacy & Data Flow section addressing ClawHub security scan findings.
v2.3.1
fix: CPU model recommendation, memory detection fallback, ASS style validation, empty SRT entries, CJK slugify. Add --version flag.
v2.3.0
Enlarge subtitle fonts (EN 44pt, ZH 56pt), yellow-on-gray clean preset, prompt user for style choice, add custom .ass option
v2.2.0
Untrack test file from repo
v2.1.0
Add multi-platform metadata (Codex, OpenClaw/ClawHub, hermes, SkillsMP) and Codex agent manifest
元数据
常见问题
Yt2bb 是什么？
Use when the user wants to repurpose a YouTube video for Bilibili, add bilingual (English-Chinese) subtitles to a video, or create hardcoded subtitle version... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 181 次。
如何安装 Yt2bb？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install yt2bb」即可一键安装，无需额外配置。
Yt2bb 是免费的吗？
是的，Yt2bb 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Yt2bb 支持哪些平台？
Yt2bb 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（darwin, linux, win32）。
谁开发了 Yt2bb？
由 Agents365.ai（@agents365-ai）开发并维护，当前版本 v2.5.0。
推荐 Skills