video-transcript skill
Skill that pulls text from videos and podcasts through the Quark cloud, runs resumable batches and proofreads the result.
- 116 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Media files are uploaded to the Quark cloud, and the install script is downloaded and run through `bash`.
- The skill runs Python scripts, `ffmpeg` and a third-party skill with access to your Quark account.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/video-transcriptDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Media files are uploaded to the Quark cloud, and the install script is downloaded and run through `bash`.
- The skill runs Python scripts, `ffmpeg` and a third-party skill with access to your Quark account.
This is third-party code. Review the repository files before installing.
What it does
video-transcript takes a video or podcast link, a local file or a file already stored in Quark and returns a transcript for review. The neighbouring `video-download` skill fetches the video, the media goes to the Quark cloud, and the official `quarkclouddrive` skill is asked for the text of each file separately. Every file gets a permanent ID, and answers, attempts and statuses are written to disk, so a batch resumes after an interruption. The agent checks whether the text is too short or only a summary, prepares a draft and fixes obvious errors with small patches while keeping the original.
Who it is for. Editors and researchers with a Quark account and a pile of recordings, such as interviews, podcasts and lectures, who need them as text.
Good fit when
- A transcript of a podcast or interview is needed from a link.
- A dozen recordings must be transcribed as a batch and resumed after a failure.
- The file is already in Quark and an ID is enough to select it.
- A draft transcript must keep the original next to the edits.
Not a fit when
- You need speaker labels, timestamps or SRT: the skill does not produce them.
- The recording must not be uploaded to a third-party cloud.
- Nobody plans to create a Quark account.
- Transcription has to run locally, without internet.
Example request
Transcribe three interview recordings from `~/Interviews` as a batch and show the status of each.Limitations
The skill does not work without the separately installed and authorised official `quarkclouddrive` skill, which the repository does not contain. It also needs macOS, Python 3.9 or newer, Node.js, `ffmpeg` and the `video-download` skill. Media is uploaded to your Quark cloud and is not deleted automatically. The cloud returns text without a guarantee of completeness or accuracy, and there are no speaker labels, timestamps or SRT. The `--doctor` check confirms that tools exist, not that the cloud works. The documentation is written in Chinese.
How to disable. Delete the install folder, `~/.claude/skills/video-transcript` by default. Media uploaded to Quark, answers and statuses in `outputs` are removed separately: the skill deletes nothing on its own.
FAQ
How do I install video-transcript?
Run `bootstrap.sh` from the repository through `bash`, as the README shows. Without `VIDEO_TRANSCRIPT_TARGET` it installs the skill into `~/.claude/skills/video-transcript`; dependencies are installed later by `install.sh`.
Is the Quark skill required separately?
Yes. You install and authorise the official `quarkclouddrive` yourself; the repository contains neither the Quark CLI nor credentials.
Does the skill give timestamps and SRT?
No. The current version builds no timestamps, SRT or speaker labels. Local FunASR transcription stayed in the old version, described in `docs/legacy-README.md`.
How do I resume an interrupted batch?
Run `transcript.py` with `--resume` and the manifest path, and `--status` shows file states. Statuses such as `auth_required` are handled by `references/quark-workflow.md`.
What does the text_received status mean?
It only means the text came back from the cloud. Completeness and accuracy stay unverified, and the agent checks them before proofreading.
Discussion
No questions or reviews yet.
Related
SkillsA skill that turns a topic into a narrated 4K video: research, script, local TTS, Remotion composition and an FFmpeg mix
A skill that turns a coding agent into a studio for voiceover explainer videos: word-level sync, a motion library and Remotion assembly
Skill that turns a video into tall subtitle collages: crops burned-in subtitles or draws reviewed lines onto real frames
A skill that pulls highlights from a long video: download, transcript, segmentation, analysis and clips with titles
The video-explainer skill turns a script into a vertical 9:16 explainer: narration, captions, Remotion and a pre-render review
Twelve skills teach an agent to read YouTube through the TranscriptOut service: transcripts, search and channels in one HTTPS call
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| video-transcript (this card) | no | 116 | ||
| Whisper | no | 112,049 | transcribes audio locally with a Whisper model; video-transcript sends the file to the Quark cloud and takes back the finished text | |
| YouTube Transcript Server | no | not required | 603 | returns a YouTube video transcript by link; video-transcript handles links, local files and batches of recordings |
README in short
The Quark cloud version replaced local FunASR: the file is uploaded, text is requested one file at a time, and checking and proofreading happen on the agent side. Details and the recovery procedure live in `SKILL.md` and `references/quark-workflow.md`.
README badge
Original SKILL.md
--- name: video-transcript description: > 视频与播客文字提取：下载音视频后调用夸克网盘 Skill 云端取稿，再由模型轻校对。 支持单条和批次、已上传网盘文件、持久对应关系与中断恢复。用于“逐字稿、转文字、提取文案” 或附音视频链接的提取请求。默认不运行本地 ASR；不承诺说话人分离、句级时间轴或 SRT。 metadata: display-name: 视频文案提取 --- # 视频文案提取：夸克云端版 链接 / 本地媒体 / 已指定的网盘文件 → 登记 → 下载或匹配 → 上传并核对 → 单文件问答取正文 → 质量检查 → 模型 patch 轻校对。 ## 入口与依赖 使用**本次读取的这份 SKILL.md 所在目录**作为 `VT_HOME`。不要按 WorkBuddy/Codex 等优先级另找一份 video-transcript；不能把其他副本当成这份的运行目录。 `VT_PY` 使用可运行脚本的 Python 3.9+，不必寻找装有 FunASR 的解释器。 首次执行先读已安装 `quarkclouddrive/SKILL.md` 及本次涉及的 `references/file-upload.md`、`references/file-search.md`、`references/assistant.md`。 下载沿用 video-download；平台解析所需登录态仍按该 Skill 处理。夸克 CLI 是外部依赖，不复制其源码或凭据。本脚本会在每次调用夸克命令前运行官方环境检查，并使用同一批次会话 ID。 ```bash python3 "$VT_HOME/scripts/transcript.py" --doctor ``` 该体检只检查本地工具与依赖位置，不代表授权或云端链路成功。缺少夸克 Skill 时按其安装说明安装；不要运行旧版 ASR 安装步骤。需要安装下载依赖才运行本 Skill 的 install.sh；已有 .env、outputs 和模型缓存保留。 ## 单条与批次 用户明确选择此 Skill 的云端工作流，即按任务范围上传指定媒体。用户要求只下载时转交 video-download。 保存用户本轮原始请求为临时 UTF-8 文件，通过 `--session-input-file` 传入；这是夸克要求的服务质量追踪参数，不应塞入额外对话或凭据。 ```bash python3 "$VT_HOME/scripts/transcript.py" "<链接或本地路径>" --session-input-file "<原始请求文件>" --wait 1800 python3 "$VT_HOME/scripts/transcript.py" --batch "<inputs.json>" --session-input-file "<原始请求文件>" --wait 1800 ``` 等待值是本次进程的复查预算，不是预计完成时间；单次网络请求可能在预算结束后才返回。 批次 JSON 是输入字符串数组，或包含 `input`、可选 `title` 的对象数组。 截图 PDF 保留 `--keep-video` 兼容参数；新版媒体默认持久保存，不自动删本地文件或云端文件。 已上传网盘文件的选定、参数、缓存与错误恢复见 [references/quark-workflow.md](references/quark-workflow.md)。 ## 状态和中断恢复