Boson Video

Install

Works with Claude Code, Codex, VS Code, Cursor, Windsurf, Claude Desktop. Read the repository before you give any tool access to your data.

npx agentmarket install connector/linboxin-boson-video

About

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list, install config and code. Checked Oct 25, 2026.

Tools

5 read · 0 write · 0 delete · 4 not yet classified

Secrets

1 (API keys, tokens or passwords)

Install scripts

None found

Runs as

Python package

Tools (9)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• page_document

  For the video's page in the chat, not for the assistant: the document as JSON, or only its version when it hasn't changed since `since`.

• page_picture

  For the video's page in the chat, not for the assistant: the picture at second `t`.

• page_sheet

  For the video's page in the chat, not for the assistant: thumbnail sheet `index`.

• video_check

  Check a claim about the video against what was said and shown at the moments it rests on (`at`, times like ["4:26"]) and the passages around them. A number that appears only on screen is checked against that moment. With a Jev key, Jev judges the meaning and code checks every number; without one, only the numbers are checked. Says which checked it.

• video_frames

  Look at the video. Either exact moments, `at=["4:26", "5:10"]`, or the new visuals the scene map found between `start` and `end` (new scenes and the build steps of slides and diagrams; repeats of an earlier picture are skipped). `limit` frames (default 6, at most 12; more than 6 come a little smaller), at full resolution when they can be fetched, each citing the moment and what was said around then.

• video_list

  The videos opened so far: key, title, channel, length, language.

• video_open

  Open a video and get its briefing. `video` is a YouTube link or id, or a local file path. The first time, this starts building it in the background (the scene map takes about 2 s; transcription about 10 s for 10 minutes and 25 s for 35 minutes) and returns as soon as the map is ready, saying when the words will be. Call it again later for the full briefing.

• video_read

  Read the transcript between `start` and `end` (times like "4:26" or "1:02:03"; no end = to the end of the video). Each spoken line is "[m:ss] original // English". A picture change is a moment on its own line, "[m:ss] state|delta|trajectory|seek: the lines new on screen"; video_frames with at=["m:ss"] shows its frame. `lang`: "both" (default), "orig" or "en". Long stretches stop at about 12k tokens and say how to continue.

• video_search

  Find where something is said or shown, in one video (`video`) or in every video opened so far. A hit on the picture cites the same moment as video_read. With a Jev key and one video, Jev's pick of the answering passage comes first.

Signals

Stars

18

Forks

0

Open issues

0

Weekly downloads

69

Last commit

Oct 15, 2026 (13 days ago)

Last release

Oct 10, 2026 (18 days ago)

Last package publish

Oct 15, 2026 (13 days ago)

From GitHub (fetched Oct 24, 2026) and the package registry.

Details

Registry name

io.github.linboxin/boson-video

Version

0.3.0

Auth

API key

Packages

pypi boson-video

License

MIT

Repository

github.com/linboxin/Boson-Video

First seen

Oct 24, 2026

Last checked

Oct 24, 2026

Source

Official MCP Registrygithub-graphql