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