Skip to content

Releases: ufoooo2025/video-context

Release list

Video Context 0.1.0a3 — English documentation

Video Context 0.1.0a3 — English documentation Pre-release
Pre-release

Choose a tag to compare

View all tags
@ufoooo2025 ufoooo2025 released this 05 Sep 21:04
v0.1.0a3
3059645

Let your AI assistant summarize videos from YouTube, Bilibili, and Xiaohongshu. Send a link to get the main points, take notes, or ask questions about the video. Xiaohongshu support is experimental.

What changed

  • English is now the primary language for the README, usage guide, contributing guide, and security and validation pages.
  • Each page includes a link to its Chinese translation.
  • The introduction explains what you can do before describing subtitles, local transcription, and MCP.

This is a documentation update. Extraction behavior is unchanged from 0.1.0a2; some runtime messages and Skill instructions remain in Chinese.

Get started

Windows users: download video-context-0.1.0a3-source.zip below, extract it, and follow the README to prepare uv, Codex CLI, and Node.js for YouTube. Run install.cmd, then reopen Codex and send a video link. First setup downloads dependencies and a speech model of about 500 MB.

English documentation · 中文文档

Validation and limits

Package privacy checks, local documentation links, and installation/version checks passed. Extraction code and tests match the previous release, whose 29 tests passed in an isolated environment. No runtime change was made in this release.

This is still a development preview. Fresh operating-system installation and real-machine macOS/Linux validation remain pending. Video Context uses subtitles and audio; it does not interpret everything shown on screen. Not every platform link is accessible. Use it locally, not as a public service accepting arbitrary URLs.

MIT licensed. Downloads exclude private samples, credentials, models, and machine settings. Use SHA256SUMS.txt to verify the files.

Loading

Video Context 0.1.0a2 · 开发预览

Video Context 0.1.0a2 · 开发预览 Pre-release
Pre-release

Choose a tag to compare

View all tags
@ufoooo2025 ufoooo2025 released this 05 Sep 20:57
v0.1.0a2
c4e8db9

Video Context 的首个公开预览版。把 YouTube、B站或小红书视频链接交给 AI 工具，先获取带时间戳的文字，再进行总结和问答。

开始使用

Windows 用户请下载下方的 video-context-0.1.0a2-source.zip，解压后按 README 准备 uv 和 Codex CLI，再双击 install.cmd。首次安装需要联网下载依赖和约 500 MB 的语音模型。安装成功后重新打开 Codex，即可尝试发送视频链接。

这一版包含什么

  • 优先使用平台字幕，没有合适字幕时在本机识别语音。
  • Markdown / JSON 导出、本地缓存，以及 MCP 和 Skill 接入。
  • B站分 P 选择、后台转写和取消、长文字分页读取。
  • MIT 协议，可商用、修改和分发，需保留版权及许可声明。

验证与限制

独立环境中的 29 项测试通过，包含实际 MCP 标准输入输出通信。YouTube 字幕、B站与小红书样本转写已有本机验证；源码安装器复用现有模型的流程也已通过。

这是开发预览版。小红书为实验性支持；首次使用仍需要准备环境。全新操作系统首次安装及 macOS/Linux 实机验证尚未完成。工具读取字幕和声音，不具备完整画面理解能力，也不保证每个链接均可提取。请勿直接用作接收任意链接的公网服务。

发布包不含私人样本、Cookie、访问参数、模型或本机配置。SHA256SUMS.txt 用于核对下载文件。

Loading