Video Transcribe
wendy7756/AI-Video-Transcriber
Transcribe and summarize a video or podcast from a URL (YouTube, TikTok, Bilibili, Apple Podcasts, SoundCloud, 30+ platforms) or from a local media/.txt file.
Extract Bilibili videos and opus/article posts into readable Markdown knowledge notes.
$ npx skills add Rimagination/bili-note --skill bili-note -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install Rimagination/bili-note bili-note --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
Claude Code skills documentation · loads skills from .claude/skills/
Install the "bili-note" agent skill from https://github.com/Rimagination/bili-note/tree/main into .claude/skills/bili-note/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bili-note", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add Rimagination/bili-note --skill bili-note -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install Rimagination/bili-note bili-note --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "bili-note" agent skill from https://github.com/Rimagination/bili-note/tree/main into .agents/skills/bili-note/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bili-note", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add Rimagination/bili-note --skill bili-note -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install Rimagination/bili-note bili-note --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "bili-note" agent skill from https://github.com/Rimagination/bili-note/tree/main into .cursor/skills/bili-note/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bili-note", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add Rimagination/bili-note --skill bili-note -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install Rimagination/bili-note bili-note --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "bili-note" agent skill from https://github.com/Rimagination/bili-note/tree/main into .gemini/skills/bili-note/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bili-note", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install Rimagination/bili-note bili-noteInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add Rimagination/bili-note --skill bili-note -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "bili-note" agent skill from https://github.com/Rimagination/bili-note/tree/main into .github/skills/bili-note/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bili-note", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add Rimagination/bili-note --skill bili-note -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install Rimagination/bili-note bili-note --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "bili-note" agent skill from https://github.com/Rimagination/bili-note/tree/main into .opencode/skills/bili-note/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "bili-note", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
bili-noteExtract Bilibili videos and opus/article posts into readable Markdown knowledge notes.
Bili Note is an agent skill from Rimagination/bili-note. Extract Bilibili videos and opus/article posts into readable Markdown knowledge notes. Use when the user asks to 提取/提炼/总结/整理 B站 or bilibili video/图文/动态/opus content, save a B站 note to a local knowledge base, include useful comments, handle multi-part videos, fetch B站 AI subtitles, download opus images, or fall back to audio transcription.
Its SKILL.md is about 3.6k tokens, which your agent loads only when the skill is triggered. The skill folder holds 32 other files, including scripts, reference files and assets (for example `README.md`, `agents/openai.yaml` and `references/bilibili-api-notes.md`).
It sits in Media & Creative, covering Transcription. It works with Bilibili and Qwen. The repository describes itself as: Extract Bilibili videos into learning-oriented Markdown notes with full subtitle and comment archives. The licence is MIT.
9 steps, taken from the step headings in SKILL.md.
Read from SKILL.md and the folder at commit 08fe812. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Ships 10 files in scripts/ (Python, from the files we listed), which the agent can run.
Shell commands in SKILL.md call:
curlpythonFrom the folder's file list and the shell code blocks in SKILL.md.
Hosts in commands or code, which the agent is likely to contact:
bilibili.comFrom URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Bili Note loads about 3.6k tokens when it runs, and up to ~4.6k if it reads all its reference files. Until then it costs about 88 tokens; SKILL.md has 630 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
The full file from Rimagination/bili-note at commit 08fe812, republished under its MIT licence (© Rimagination). 630 words, ~3,646 tokens.
.claude/skills/bili-note/SKILL.md (or your agent's skills folder). This skill also uses 28 other files; get the full folder from GitHub.把 B站视频和图文动态变成可检索、可复用的 Markdown 知识笔记。视频优先拿字幕;字幕拿不到时再转写音频;图文优先抓正文、图片和代码块;用户要评论区时抓取评论并过滤无关讨论。
联网或登录态操作必须先使用 web-access。
网页 AI 字幕支持两条路线:Chrome 使用 web-access,Edge 使用开启远程调试的 Chromium DevTools Protocol 直连。两条路线都让已登录的 B站页面自己请求字幕接口,不读取或复制 Cookie/profile。Playwright 临时浏览器不属于受支持路线。
Bili Note 与 DyNote 共享可复用本地资源。默认共享目录是 %USERPROFILE%\.cache\rimagination-notes,Qwen3-ASR 环境默认是 %USERPROFILE%\.cache\rimagination-notes\qwen3-asr-venv。如果任一 skill 已经安装过 Qwen3-ASR,另一个 skill 必须优先复用,不要重复安装。Hugging Face、Whisper 和 faster-whisper 缓存按本机通用缓存复用。
B 站视频优先拿字幕,但字幕/转写不等于完整理解。长视频如果字幕字数明显偏少,可能说明核心信息在画面、PPT、板书、代码演示、屏幕操作或无解说片段中。
on 或 off 后才能进入后续流程。询问时说明:on 会抽取最多 12 帧,先合成为一张 4×3 联系图,再按需要打开单帧,视觉输入会增加 tokens;off 只走字幕、转写、评论和元数据路线,节省视觉输入 tokens。on 时,给 run_bili_note.py 传 --visual-review on;选择 off 时传 --visual-review off。脚本默认 --visual-review ask,直接在终端运行也会在轻量预检后停下来等待选择。metadata/note_budget.json。如果 visual_dependency.needs_visual_review=true,不要把稀疏字幕写成完整学习笔记。visual_dependency.warnings,并说明是否已经补了视觉证据。关键帧视觉理解采用低成本的两阶段读取:先看一张 4×3 联系图建立全片画面地图;联系图中出现文字、代码、图表或界面细节时,再依据工作目录的 keyframes_manifest.json 打开对应单帧。归档后对应路径是 archive_dir/metadata/keyframes_manifest.json。不要默认把 12 张完整图片全部送入模型。
选择 on 后,先查看 keyframes/contact_sheet.png,再结合 keyframes/README.md 和工作目录的 keyframes_manifest.json 定位时间点。把视觉观察写入材料包的 metadata/visual_review.md,每条关联 KF-Pxx-xx、分P、时间和观察结论;不确定的画面标记为待核实。关键帧证据可通过 indexes/关键帧索引.jsonl 和总证据索引回查。
默认路线尽量零第三方 Python 依赖:视频元数据、公开字幕、图文正文、图片清单、评论、归档、证据索引和笔记预算都用标准库完成。网页 AI 字幕和音频转写是增强路线,不是启动门槛。
第一次使用、换机器、用户怀疑依赖不全,或准备使用网页 AI 字幕 / 音频转写兜底时,先运行:
$skill = "$env:USERPROFILE\.codex\skills\bili-note"
$py = "python"
& $py "$skill\scripts\check_environment.py"根据检查结果选择路线:
public_subtitles_comments_archive=OK:优先走默认字幕/图文、评论和归档流程。browser_ai_subtitles=OK:当公开接口只有 ai-zh 且 subtitle_url 为空时,走 Chrome + web-access 网页 AI 字幕。edge_browser_ai_subtitles=OK:需要 Edge 远程调试端口和 websocket-client,总入口加 --browser edge。audio_asr_fallback=OK:只有字幕和网页 AI 字幕都不可得、且用户确实需要完整转写时,才走音频转写。中文或未指定语言优先共享 Qwen3-ASR;明确外语视频优先 Whisper 系后端。web-access 页面或 Edge 远程调试页面使用;不要读取或复制 Cookie/profile,不要强制结束用户浏览器进程。两种浏览器路线都不可用时就跳过网页 AI 字幕并说明覆盖范围。check_environment.py 判断当前可走路线。--visual-review on 或 --visual-review off 传给总入口。run_bili_note.py 一键完成可自动化部分。它会自动识别 /video/BV...、/opus/...、/dynamic/... 或纯 opus id。extract_bilibili_opus.py 路线:抓正文、标题、作者、发布时间、图片、代码块、图文证据索引;用户加 --comments 时抓图文评论。--download-subtitles。ai-zh 但 subtitle_url 为空,不要说“没有字幕”;改走“网页 AI 字幕”流程。--comments 抓取主评论和子评论;写入笔记时过滤打卡、求资料、广告、闲聊等技术无关内容。on 时,在写前定标前查看 keyframes/contact_sheet.png,必要时查看对应单帧,并保存 metadata/visual_review.md。metadata/note_budget.json,把推荐字数区间、压缩比目标、写作粒度、互动质量倍率、证据块数量和 visual_dependency 作为本次笔记的写作目标。视频按时长、字幕字数、证据块、评论量和互动质量定标;图文按正文长度、图片/代码/证据块、评论量和互动质量定标。visual_dependency.risk 是 medium 或 high,先补关键帧/OCR/多模态视觉理解,或明确告诉用户当前缺少视觉理解能力,不能写成完整解析。[1][2],不直接堆长证据 ID。score_bili_note.py 校验笔记字数、压缩比、每分钟/每篇笔记密度和证据引用比例。评分只做 QA 和微调,不代替写前定标;太短时优先补“学习收获、知识地图、概念卡、实战流程、坑点、自测题”,不要只堆分P摘要或段落摘要。在 PowerShell 中先设定 skill 路径:
$skill = "$env:USERPROFILE\.codex\skills\bili-note"
$py = "python"& $py "$skill\scripts\check_environment.py"需要给其他脚本读取时输出 JSON:
& $py "$skill\scripts\check_environment.py" --json默认先用总入口。它会自动跳过已有输出,适合断点续跑。
& $py "$skill\scripts\run_bili_note.py" "https://www.bilibili.com/video/BVxxxx/" `
--work-dir ".\tmp_bili_extract" `
--archive-dir "D:\knowledge\知识库\Rag技术\原始材料\BVxxxx_视频短标题" `
--visual-review on `
--comments如果用户选择关闭,把上面的 --visual-review on 改为 --visual-review off。每次视频运行都重新询问,不能沿用上一次选择。
图文/动态也用同一个入口:
& $py "$skill\scripts\run_bili_note.py" "https://www.bilibili.com/opus/1194341967364882439" `
--work-dir ".\tmp_bili_opus" `
--archive-dir "D:\knowledge\知识库\Rag技术\原始材料\O1194341967364882439_图文短标题" `
--comments如果只想保存图文正文和图片 URL,不下载图片文件,加:
--no-download-images如果普通接口没有字幕,但网页播放器能拿到 AI 字幕,Chrome 先用 web-access 打开已登录页面并取得 target id,然后加:
--browser-target "CDP_TARGET_ID"运行后先看:
bili_note_run_report.md:本次跑了什么、跳过了什么、下一步读哪里。visual_preflight.json:询问前的时长、字幕密度、风险和建议,方便复核本次选择依据。keyframes/contact_sheet.png:开启关键帧视觉理解时生成的 4×3 联系图。keyframes_manifest.json:工作目录中的关键帧编号、分P、时间和单帧路径。archive_dir/metadata/keyframes_manifest.json:归档后的关键帧清单。archive_dir/metadata/visual_review.md:视觉模型对关键帧的观察结果,由 Agent 在理解后写入。archive_dir/indexes/证据索引.jsonl:写总结时可引用的图文/字幕/评论证据块。archive_dir/indexes/图文全集.md:完整图文正文合集。archive_dir/indexes/字幕全集.md:完整字幕合集。archive_dir/metadata/note_budget.json:推荐笔记字数、压缩比、写作粒度和字幕稀疏时的画面依赖提示。archive_dir/metadata/visual_preflight.json:归档后的关键帧选择预检记录。& $py "$skill\scripts\extract_bilibili.py" "BVxxxx" --out ".\tmp_bili_extract"& $py "$skill\scripts\extract_bilibili.py" "BVxxxx" --out ".\tmp_bili_extract" --parts all --download-subtitles当 subtitle_probe.json 里有 ai-zh,但 subtitle_url 为空时,使用这条路线。Chrome 需要 web-access 的 /targets 和 /eval 代理接口;Edge 使用直接 CDP 路线。
Chrome 路线:
web-access 打开已登录 Chrome 中的 B站视频页,确认页面已加载。curl.exe -s http://localhost:3456/targets& $py "$skill\scripts\fetch_browser_ai_subtitles.py" --target "CDP_TARGET_ID" --out ".\tmp_bili_extract"Edge 路线:
--remote-debugging-port=9222 启动 Edge 独立实例,在其中登录 B 站并打开视频页。python -m pip install websocket-client。& $py "$skill\scripts\run_bili_note.py" "BVxxxx" `
--work-dir ".\tmp_bili_extract" `
--browser edge `
--subtitle-mode browser也可以直接运行字幕脚本;它会优先选择 Edge 中 URL 包含 bilibili.com 的页面:
& $py "$skill\scripts\fetch_browser_ai_subtitles.py" `
--browser edge `
--edge-cdp-url "http://127.0.0.1:9222" `
--out ".\tmp_bili_extract"输出包括:
browser_ai_subtitle_urls.json:每个分P的 AI 字幕 URL。browser_ai_subtitle_manifest.json:下载清单。browser_ai_subtitles\*.txt:纯文本字幕。browser_ai_subtitles\*.srt:SRT 字幕。browser_ai_subtitles\*.subtitle.json:B站原始字幕 JSON。这条路线让 B站页面自己用登录态请求 /x/player/wbi/v2,不读取、不打印浏览器 cookie。
如果下载到的 AI 字幕明显乱码、内容与标题或课程主题不相干,不要把它当作可靠全文。优先标注“AI 字幕质量不可用”,再考虑 ASR 兜底;若 ASR 也不可用,只能基于分P标题、简介、评论和元数据生成有限学习指南,并明确局限。
& $py "$skill\scripts\extract_bilibili.py" "BVxxxx" --out ".\tmp_bili_extract" --comments图文评论用总入口或图文脚本:
& $py "$skill\scripts\extract_bilibili_opus.py" "https://www.bilibili.com/opus/1194341967364882439" --out ".\tmp_bili_opus" --comments抓取后检查 comments.md 和 comments_raw.json。写入最终笔记时,只保留与内容主题相关的评论、纠错、技术补充、实践经验和有价值问题。
只有在字幕和网页 AI 字幕都不可用时才用本地自动语音识别。长视频不要默认全量转写,除非用户明确要求。中文或未指定语言默认 --asr-backend auto 会优先共享 Qwen3-ASR;明确外语视频时用 Whisper 系后端。
& $py "$skill\scripts\extract_bilibili.py" "BVxxxx" --out ".\tmp_bili_extract" --parts "1,10,38" --download-audio --transcribe --asr-backend auto --asr-model base首次使用 Qwen3-ASR 前先安装共享环境:
& $py "$skill\scripts\setup_qwen_asr_env.py"
& $py "$skill\scripts\check_environment.py"中文视频可显式指定 Qwen3-ASR:
& $py "$skill\scripts\extract_bilibili.py" "BVxxxx" --out ".\tmp_bili_extract" --parts "1,10,38" --download-audio --transcribe --asr-backend qwen3-asr把临时提取目录整理成长期材料包。这个步骤默认要做,方便后续根据总结继续提问和回查证据。
& $py "$skill\scripts\archive_bili_materials.py" --extract-dir ".\tmp_bili_extract" --archive-dir "D:\knowledge\知识库\Rag技术\原始材料\BVxxxx_视频短标题"长期材料包包含:
articles/图文全文.md:完整图文 Markdown。articles/图文全文.txt:完整图文纯文本。images/:图文图片和图片清单。indexes/图文全集.md:合并后的完整图文正文。indexes/图文全集.jsonl:按图文内容块切分,适合检索和问答。indexes/图文证据索引.md:图文证据块,适合给总结加引用。indexes/图文证据索引.jsonl:图文证据块,适合程序检索。subtitles/txt/:每个分P的完整纯文本字幕。subtitles/srt/:每个分P的完整 SRT 字幕和时间轴。subtitles/json/:B站原始字幕 JSON。comments/comments_raw.json:完整评论原始结构。comments/评论全集.md:完整评论 Markdown。indexes/字幕全集.md:合并后的完整字幕,适合人工阅读。indexes/字幕全集.jsonl:按字幕片段切分,适合检索和问答。indexes/字幕证据索引.md:按时间段合并的字幕证据块,适合给总结加引用。indexes/字幕证据索引.jsonl:按时间段合并的字幕证据块,适合程序检索。indexes/评论全集.jsonl:按评论/回复切分,适合检索和问答。indexes/评论证据索引.jsonl:按评论/回复生成的证据块。indexes/证据索引.jsonl:图文/字幕证据和评论证据的合并索引。metadata/:内容元数据、字幕清单、图文清单、评论清单。metadata/note_budget.json:按内容信息量和互动质量生成的笔记预算,用于控制长短视频、长短图文的提炼粒度,并提示字幕稀疏时的视觉补证需求。写笔记前先看预算。预算是本次笔记的目标,不是写完后才补救的报告:
Get-Content -Encoding UTF8 "D:\knowledge\知识库\Rag技术\原始材料\BVxxxx_视频短标题\metadata\note_budget.json"重点先确定:
recommended_note_chars_min 到 recommended_note_chars_max。granularity 和 writing_guidance。duration_minutes、subtitle_chars,图文看 content_chars、reading_minutes_estimate。subtitle_chars_per_minute 和 visual_dependency;长视频字幕过少时先补关键帧/OCR/多模态视觉理解。quality_multiplier 和 quality_metrics,用于判断是否值得保留更多细节。all_evidence_blocks,用于决定关键判断需要覆盖多少证据。写完最终 Markdown 后,用归档目录里的预算给笔记打分:
& $py "$skill\scripts\score_bili_note.py" `
--archive-dir "D:\knowledge\知识库\Rag技术\原始材料\BVxxxx_视频短标题" `
--note-path "D:\knowledge\知识库\Rag技术\观点X:视频短标题.md" `
--out "D:\knowledge\知识库\Rag技术\原始材料\BVxxxx_视频短标题\metadata\note_score.json"也可以直接把评分结果写入笔记正文:
& $py "$skill\scripts\update_note_budget_section.py" `
--archive-dir "D:\knowledge\知识库\Rag技术\原始材料\BVxxxx_视频短标题" `
--note-path "D:\knowledge\知识库\Rag技术\观点X:视频短标题.md"重点看:
status:too_short 表示遗漏风险高,too_long 表示可能重复堆料,ok 表示落在写前推荐区间。quality_multiplier:点赞、收藏、投币、评论、弹幕、分享、播放量和发布距今天数形成的互动质量倍率。actual_compression_ratio:笔记字数 / 字幕字数。长视频应允许更高总字数,但仍要保持压缩。note_chars_per_minute:每分钟视频对应多少笔记字。长课程不应和短视频接近同一个总字数。evidence_reference_ratio:笔记中引用的证据块占比。关键判断要能回查图文证据 O...、字幕证据 Pxx@... 或评论证据 C...;正文用统一数字编号,文末编号脚注和 reference-style 链接定义必须指向含完整证据 ID 的归档文件。默认输出必须是“学习型笔记”,不是目录搬运、分P流水账或证据清单。读完后应有“我真的学会了一些东西”的获得感。
# 标题## 学完你应该获得什么:用 5-8 条写清楚读者学完后能理解、判断或完成什么。## 一句话总论:说明视频最核心的判断,避免只复述标题。## 适用场景与前置知识:这套内容适合谁、不适合谁,读者需要先知道什么。## 知识地图:把核心概念、模块、流程和它们之间的关系讲清楚。课程型视频要有模块表;短观点视频要有论证链。## 核心概念卡:每个重要概念都写成“是什么、为什么重要、怎么用、常见误区、相关证据”。## 方法或流程:把原内容中的操作、架构、决策流程、参数选择、评估方法抽成可复用步骤。## 关键洞察:写作者真正想表达的判断,区分“事实描述、经验判断、推荐做法、限制条件”。## 实践清单:给出可以照着做的步骤、检查项、失败信号和排错方向。## 坑点与反例:整理视频和评论区提到的踩坑、争议、误区、边界条件。## 自测题:写 5-10 个问题和简短答案,帮助人或 Agent 检查是否学会。## 证据与原文位置:只放关键证据引用,不要让证据淹没学习内容。## 来源、覆盖与局限:URL、BVID、UP、发布时间、字幕/评论覆盖、原始材料路径和局限放在后面。[1][2];图文证据、字幕证据、评论证据共用一套编号,按正文首次出现顺序递增。不要在正文段落里直接塞很长的 O<opus_id>-E001、Pxx@hh:mm:ss-hh:mm:ss 或 C<rpid>。## 来源、覆盖与局限 之前放 ## 证据脚注。脚注用有序列表写明编号和证据链接,例如 1. [图文证据 E006](原始材料/O.../indexes/图文证据索引.md#O...-E006);列表后再放 reference-style 链接定义,例如 [1]: 原始材料/O.../indexes/图文证据索引.md#O...-E006 "图文证据 E006",让正文里的 [1] 也能点击跳转。## 证据与原文位置 作为证据总览,但总览里也只写 [1]、[2] 这样的编号,不再直接显示长证据 ID。note_budget.json 定标,再开始写笔记;写后评分只用于验收和微调,不要把主要长度决策推迟到评分阶段。不要把“只看了标题/目录”的内容写成“完整提取”。如果只抓到部分字幕或只转写了部分分P,明确列出覆盖范围。
/x/player/v2 可能只返回 ai-zh 字幕元信息,subtitle_url 为空。/x/player/wbi/v2 常能返回真正的 AI 字幕 URL。window.__INITIAL_STATE__ 读取;公开 polymer 动态接口可能返回风控错误。basic.comment_type 和 basic.comment_id_str;抓评论时不要硬套视频的 type=1/oid=aid。images/ 或 images_manifest.json;仅凭图片 URL 不足以判断内容时,应说明图片视觉内容未被完整理解。scripts/check_environment.py:检查核心流程、网页 AI 字幕、音频转写和测试依赖是否可用。scripts/setup_qwen_asr_env.py:创建/更新共享的 Qwen3-ASR Python 环境。scripts/run_qwen_asr.py:调用 Qwen3-ASR-0.6B,可按 chunk 分段避免显存溢出。scripts/run_bili_note.py:一键运行视频/图文提取、评论、归档和证据索引流程。scripts/extract_bilibili.py:元数据、字幕探测、普通字幕、音频、音频转写、评论抓取。scripts/extract_bilibili_opus.py:B站图文/动态正文、图片、代码块和图文评论抓取。scripts/fetch_browser_ai_subtitles.py:通过已登录网页播放器下载 B站 AI 字幕。scripts/edge_cdp.py:连接 Edge 远程调试页面的可选 CDP 适配器。scripts/extract_video_keyframes.py:按分P时长抽取最多 12 张代表帧,生成 4×3 联系图和关键帧清单。scripts/archive_bili_materials.py:归档完整材料,生成全文索引、证据索引和带字幕密度/视觉依赖提示的笔记预算。scripts/score_bili_note.py:按 metadata/note_budget.json 验收最终笔记的长度、压缩比、证据引用和视觉依赖提示。scripts/update_note_budget_section.py:把预算、互动质量和信噪比评分写回 Markdown 笔记。references/bilibili-api-notes.md:接口细节和已知坑。© Rimagination, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
SKILL.md and 28 other files (scripts, references, assets) in the repository root of Rimagination/bili-note.
Open the folder on GitHubat commit 08fe812
Bili Note next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Bili Note this skillRimagination/bili-note | 320 | — | ~3.6k | Automated safety check: Pass | MIT | |
| Video Transcribewendy7756/AI-Video-Transcriber | 3.3k | — | ~937 | Automated safety check: Notes | Apache-2.0 | |
| Bilibili Transcribechubbyguan/chubbyskills | 1.2k | 1 repos | ~578 | Automated safety check: Notes | MIT | |
| Video To Subtitle Summaryimlewc/video-to-subtitle-summary-skill | 212 | — | ~4.6k | Automated safety check: Notes | MIT | |
| Video To NotesKIRVO-REPORTING/video-to-notes | 105 | — | ~1.5k | Automated safety check: Pass | MIT | |
| Hotclipxixihhhh/hotclip | 299 | — | ~1.1k | Automated safety check: Pass | AGPL-3.0 |
wendy7756/AI-Video-Transcriber
Transcribe and summarize a video or podcast from a URL (YouTube, TikTok, Bilibili, Apple Podcasts, SoundCloud, 30+ platforms) or from a local media/.txt file.
chubbyguan/chubbyskills
哔哩哔哩视频 → 下载 → 转录 → 存为 Markdown 的完整工作流. An agent skill from chubbyguan/chubbyskills.
imlewc/video-to-subtitle-summary-skill
A skill your agent uses when user provides a short video platform URL or local video/audio file and wants subtitles/AI summary, or when user asks to list their own AI Douyin historical tasks.
KIRVO-REPORTING/video-to-notes
Use immediately for any bare YouTube or YouTube Shorts URL, youtu.be link, Bilibili or b23.tv link, or other video URL; do not ask what the user wants.
xixihhhh/hotclip
Turn long videos & livestream VODs into viral vertical shorts, 100% locally — on-device transcription, LLM highlight detection, 9:16 reframe with karaoke captions, and a per-clip render-QA report.
chubbyguan/chubbyskills
学习笔记自动化:视频/播客转录 → 知识点提取 → 闪卡生成 → 知识图谱更新。触发词:学习笔记、闪卡、Anki、知识提取、视频学习
Categories
Extract Bilibili videos and opus/article posts into readable Markdown knowledge notes. Bili Note is an agent skill from Rimagination/bili-note. Extract Bilibili videos and opus/article posts into readable Markdown knowledge notes.
Bili Note fits situations like: the user asks to 提取/提炼/总结/整理 B站; bilibili video/图文/动态/opus content; save a B站 note to a local knowledge base; include useful comments.
Run `npx skills add Rimagination/bili-note --skill bili-note -a claude-code`. Or copy the skill folder (the Rimagination/bili-note repository) into .claude/skills/bili-note in your project. Claude Code loads it when a task matches its description.
Run `npx skills add Rimagination/bili-note --skill bili-note -a codex`. Or copy the skill folder (the Rimagination/bili-note repository) into .agents/skills/bili-note in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add Rimagination/bili-note --skill bili-note -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/bili-note, .gemini/skills/bili-note, .github/skills/bili-note and .opencode/skills/bili-note in your project.
Going by SKILL.md and its folder, Bili Note needs Python for the scripts in its folder and the command-line tools its instructions call (curl and python). Our summary lists: Python 3.
SKILL.md names 1 domain. In commands or code: bilibili.com; the agent is likely to contact it when it follows the instructions. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
Bili Note is published under the MIT licence (from the LICENSE file in the skill folder). It allows redistribution, so the full SKILL.md is shown on this page.
About 3.6k tokens (SKILL.md is roughly 15k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 997 tokens, read only when the agent opens those files.
Skills that share tags, products or a category with Bili Note: Video Transcribe (wendy7756/AI-Video-Transcriber, 3.3k stars), Bilibili Transcribe (chubbyguan/chubbyskills, 1.2k stars), Video To Subtitle Summary (imlewc/video-to-subtitle-summary-skill, 212 stars) and Video To Notes (KIRVO-REPORTING/video-to-notes, 105 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Rimagination (a GitHub user) maintains it in Rimagination/bili-note, which has 320 GitHub stars. The repository was last updated on September 22, 2026.
Source: Rimagination/bili-note on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.