Bilibili Subtitle Downloader

下载 Bilibili 视频字幕，将其进行分块以供 LLM（大语言模型）处理，并生成高质量的总结。当用户提供 Bilibili BV 号或 URL，并希望获取视频内容的总结、核心要点或详细的分解时使用。

Security Scan

The skill largely does what it claims (download and chunk Bilibili subtitles) but has inconsistencies around where it saves login cookies/sessions and it stores sensitive cookies in plaintext — review and trust the code before use.

Purpose & Capability

Name/description (download Bilibili subtitles and chunk for LLM summarization) matches the included scripts: they log in via Bilibili QR, call Bilibili APIs, download subtitles, and write chunk files. Required capabilities (network access, local file writes) are appropriate for this purpose.

Instruction Scope

SKILL.md instructs running the provided scripts and handling QR outputs; that matches script behavior. However SKILL.md claims cookies will be saved to ~/.openclaw/workspace/bilibili_cookie.txt for both workflows — in code, download_and_chunk.py does save a cookie string to ~/.openclaw/workspace/bilibili_cookie.txt, but cheese_downloader.py instead writes a JSON session file named 'bilibili_cheese_session.json' in the current working directory. The documentation vs implementation mismatch may confuse runtime behavior.

Install Mechanism

Instruction-only skill with bundled Python scripts; there is no external download/install step in the skill manifest. The scripts depend on Python packages (bilibili_api, qrcode, aiohttp, requests) but nothing in the manifest pulls arbitrary remote code during install.

Credentials

The scripts perform QR-based login and persist authentication cookies/credentials in plaintext files: download_and_chunk.py writes a cookie string to ~/.openclaw/workspace/bilibili_cookie.txt; cheese_downloader.py writes credential JSON to a local 'bilibili_cheese_session.json'. Storing SESSDATA/bili_jct/buvid3 in plain files is sensitive because those cookies can be used to impersonate your Bilibili account. No environment variables are requested, which is consistent, but the persistent local storage of credentials is a privacy/security concern and should be documented and protected.

Persistence & Privilege

The skill does not request global 'always: true' or elevated platform privileges. It writes files to bili_temp/ and to ~/.openclaw/workspace/ (cookie), and creates a session file in cwd. Writing its own session/cookie files is expected, but be aware these files persist unless removed and could be reused by the scripts later.

What to consider before installing

This skill appears to implement the advertised subtitle-download-and-chunking workflow, but you should: - Review and understand the code before running. The scripts perform network calls to api.bilibili.com and save authentication cookies locally. - Be cautious about scanning the QR: scanning logs your account into the running environment; only do so if you trust the code and environment. - Note the inconsistency: one script saves cookies to ~/.openclaw/workspace/bilibili_cookie.txt, the other saves a JSON session to ./bilibili_cheese_session.json. That may affect subsequent runs—verify where credentials are stored and read. - Treat the saved cookie/session files as sensitive: they contain tokens (SESSDATA, bili_jct, buvid3) that can be used to impersonate your account. Consider deleting these files after use or running the skill in an isolated environment. - If you want stronger security, modify the scripts to encrypt stored credentials, store them in a secure credential store, or avoid persistent storage entirely. If you do not fully trust the source, run the scripts in a disposable VM or container, and remove the cookie/session files after use.

Bilibili 字幕下载器技能

此技能通过使用专用的 Python 脚本和子智能体 (sub-agent) 编排，自动化提取和总结 Bilibili 视频字幕的流程。

工作流程

1. 提取字幕: 运行自带的脚本来下载并分块字幕。普通视频，均为 BV 号开头

   python3 skills/bilibili-subtitle-downloader/scripts/download_and_chunk.py <BV_ID>
   

   • 登录检查: 如果脚本输出 QR_CODE_READY:<PATH>，它将等待用户扫描二维码。您应该将此图像发送给用户。
   • 保存 Cookie: 成功登录后，脚本会自动将 Cookie 保存到 ~/.openclaw/workspace/bilibili_cookie.txt。

2. 处理输出: 解析脚本输出的 RESULT_JSON，以获取分块文件列表。分块文件命名格式：

   • 普通视频 (BV号): bili_temp/<BV_ID>/<BV_ID>_chunk_0.txt
   • 课程剧集 (EP号): bili_temp/<EP_ID>/chunk_0.txt

Bilibili 课程 (Cheese) 工作流程

1. 提取课程/剧集信息: 使用课程专属脚本获取元数据和字幕。课程或者视频，往往由 SS 或者 EP 开头

   python3 skills/bilibili-subtitle-downloader/scripts/cheese_downloader.py <SS_ID or EP_ID>
   

   • 登录: 脚本将生成一个 bilibili_login_qr.png。扫描它以登录。
   • SS_ID 模式: 如果提供 SS_ID（如 ss123），脚本将打印课程信息和所有剧集列表，需要使用具体的 EP_ID 来获取字幕。
   • EP_ID 模式: 如果提供 EP_ID（如 ep456），脚本将下载字幕并切分保存到 bili_temp/ep456/ 目录，输出 RESULT_JSON。

子智能体指令

在生成用于总结的子智能体时，请使用以下提示词 (prompt) 模式：

请阅读以下 Bilibili 视频字幕分块，并提供全面、准确的总结。

要求：

• 捕获所有关键的技术细节、具体的数据点和逻辑步骤。
• 使用标题保持清晰的结构。
• 明确主旨和可执行的要点。
• 风格：专业、信息丰富且详细。

字幕文件： [PATH_TO_CHUNK]

资源

• 脚本: scripts/download_and_chunk.py - 处理 Bilibili API 交互和基于 Token 的安全分块。