bilibili-learning-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| BILIBILI_COOKIE | No | Optional Bilibili login cookie (at least SESSDATA) used to improve search reliability and enable subtitle access, especially AI subtitles. Example: SESSDATA=xxx; bili_jct=yyy |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
B站学习助手。search_videos 找视频,get_video_info 看简介和分P,get_subtitles 读字幕(做笔记和总结的最佳素材),get_hot_comments 判断视频口碑。本服务只提供数据,不调用 LLM,总结由你完成。如果工具返回 Cookie 相关的提示,请把配置方法转告用户,不要原样重复调用。
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| search_videosA | 在 B 站搜索学习类视频,返回标题、UP主、BV号、时长、播放量和链接。 |
| get_video_infoA | 读取单个 B 站视频的详细信息:标题、简介、UP主、分P列表(含 cid)、互动数据、时长。 |
| get_subtitlesA | 读取视频某一 P 的字幕,按 "[mm:ss] 文本" 逐行返回,适合做带时间戳的笔记。 |
| get_hot_commentsA | 按点赞数排序返回视频的热门评论,用来判断视频质量和观众口碑。 |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 4 tools
Each tool targets a distinct resource: discovery (search_videos), metadata (get_video_info), content (get_subtitles), and social proof (get_hot_comments). The descriptions explicitly draw boundaries, e.g. 'already know the BV id → don't use search_videos' and 'don't use get_hot_comments as a substitute for get_subtitles'.
All four names follow a consistent snake_case verb_noun pattern: search_videos, get_video_info, get_subtitles, get_hot_comments. The verb choice (search vs get) also maps meaningfully to the operation type.
Four tightly scoped tools that each earn their place in a single linear workflow (find → inspect → read → validate). Nothing feels redundant or missing at the tool-count level for a video-learning server.
The surface covers the full learning lifecycle: discover videos, get metadata/分P info, read timestamped subtitles with pagination, and check audience feedback. Minor gaps exist — no way to list a UP主's other videos, no cross-video/collection search, and no tool to persist notes — but agents can work around these.