Bilibili Video Helper

This skill implements video search, information mining, and subtitle/comment/barrage extraction through Bilibili's public APIs.

API Cookie Requirements Overview

Function	API	Cookie Required
Video Search	`/x/web-interface/search/type`	❌
Video Details	`/x/web-interface/view`	❌
Comment List	`/x/v2/reply`	❌
Subtitle List	`/x/player/v2`	✅
Barrage	`yt-dlp`	✅

Note: The three functions of search, video details, and comments do not require a Cookie and can be called directly. Subtitles and barrage require a Cookie.

Prerequisites: Cookie (Only required for subtitles and barrage)

If you need to obtain subtitles or barrage, guide users to provide a Cookie:

Please log in to Bilibili on your desktop browser, press F12 to open Developer Tools, switch to the Network tab, refresh the page and click any request, then copy the entire cookie: ... content from the Request Headers on the right and send it to me.

Save as an environment variable:

bash

COOKIE='The complete cookie content provided by the user'

Core Workflow

1. Video Search (No Cookie Required)

API:

GET /x/web-interface/search/type

bash

curl -s "https://api.bilibili.com/x/web-interface/search/type?keyword=AI&page=1&page_size=10&search_type=video" | \
  jq '.data.result[] | {title: .title, author: .author, play: .play, duration: .duration, bvid: .bvid}'

Parameter Explanation:

```
keyword
```
: Search keyword
```
page
```
: Page number (starts from 1)
```
page_size
```
: Number of items per page (max 50)
```
search_type=video
```
: Search type (video)

Return Fields:

```
title
```
: Title (contains
```
<em class="keyword">
```
highlight tags)
```
author
```
: Uploader
```
play
```
: View count
```
duration
```
: Duration (e.g., "13:28")
```
bvid
```
: Video BVID

2. Get Video Details (No Cookie Required)

API:

GET /x/web-interface/view

bash

BVID="BV1hNFdz4EZp"

curl -s "https://api.bilibili.com/x/web-interface/view?bvid=$BVID" | \
  jq '.data | {title, bvid, aid, cid, owner: .owner.name, desc, duration, stat}'

Return Fields:

Field	Explanation
`title`	Title
`bvid`	BVID
`aid`	AID
`cid`	CID of the video segment
`owner.name`	Uploader
`desc`	Description
`duration`	Duration (seconds)
`stat.view`	View count
`stat.like`	Like count
`stat.coin`	Coin count
`stat.favorite`	Favorite count
`stat.share`	Share count
`stat.danmaku`	Barrage count
`stat.reply`	Comment count

Note: All the above metadata (view count, like, coin, favorite, share, barrage, comment) can be obtained without a Cookie.

3. Get Comments (No Cookie Required)

API:

GET /x/v2/reply

Step 1: Get AID

bash

AID=$(curl -s "https://api.bilibili.com/x/web-interface/view?bvid=$BVID" | jq -r '.data.aid')

Step 2: Get Comment List

bash

curl -s "https://api.bilibili.com/x/v2/reply?type=1&oid=$AID&sort=1&ps=30" | \
  jq '.data.page, (.data.replies | length)'

Extract Popular Comments:

bash

curl -s "https://api.bilibili.com/x/v2/reply?type=1&oid=$AID&sort=1&ps=20" | \
  jq -r '.data.replies | sort_by(-.like) | .[] | "【\(.member.uname)】\(.content.message[:100])... [👍\(.like)]"'

Parameter Explanation:

```
type=1
```
: Video type
```
oid
```
: Video AID
```
sort=0
```
: Sort by time;
```
sort=1
```
: Sort by popularity;
```
sort=2
```
: Sort by reply count
```
ps
```
: Number of items per page (max 49)
```
pn
```
: Page number

4. Get Subtitles (Cookie Required)

API:

GET /x/player/v2

Step 1: Get CID

bash

CID=$(curl -s "https://api.bilibili.com/x/web-interface/view?bvid=$BVID" | jq -r '.data.cid')

Step 2: Get Subtitle List

bash

curl -s "https://api.bilibili.com/x/player/v2?bvid=$BVID&cid=$CID" \
  -H "Cookie: $COOKIE" \
  -H "Referer: https://www.bilibili.com/video/$BVID" | \
  jq '.data.subtitle.subtitles[] | {lan: .lan, lan_doc: .lan_doc, url: .subtitle_url}'

Return Example:

json

{"lan": "ai-zh", "lan_doc": "中文", "url": "//aisubtitle.hdslb.com/..."}
{"lan": "ai-en", "lan_doc": "English", "url": "//aisubtitle.hdslb.com/..."}

Step 3: Download Subtitle Content (No Cookie Required)

bash

SUBTITLE_URL="https://aisubtitle.hdslb.com/bfs/ai_subtitle/prod/xxx"
curl -s "$SUBTITLE_URL" | jq -r '.body[].content'

Subtitle Language Codes:

```
ai-zh
```
: Chinese (AI-generated)
```
ai-en
```
: English (AI-generated)
```
ai-ja
```
: Japanese (AI-generated)
```
ai-es
```
: Spanish (AI-generated)
```
ai-pt
```
: Portuguese (AI-generated)
```
ai-ar
```
: Arabic (AI-generated)

Tip: Bilibili automatically generates AI subtitles for videos. Even if the uploader hasn't uploaded subtitles, AI-generated subtitles may be available.

5. Get Barrage (Cookie Required)

The barrage API returns data in protobuf format, which is complex to parse. It is recommended to use

yt-dlp

bash

yt-dlp --add-header "Cookie:$COOKIE" --write-sub --sub-lang danmaku --skip-download -o "/tmp/%(id)s.%(ext)s" "https://www.bilibili.com/video/$BVID"

Barrage is saved in XML format. Extract plain text:

bash

cat /tmp/${BVID}.danmaku.xml | sed 's/<d /\n<d /g' | grep -oP '>[^<]+<' | tr -d '<>' | grep -v '^$'

Access Restriction Handling (412 Precondition Failed)

If you encounter a Bilibili 412 error, it means the current environment is under risk control. You need to ask the user to re-obtain the Cookie.

6. Get Uploader's Video List (No Cookie Required)

API:

GET /x/space/wbi/arc/search

bash

UID="12345678"

curl -s "https://api.bilibili.com/x/space/arc/search?mid=$UID&ps=30&pn=1&order=pubdate" \
  -H "User-Agent: Mozilla/5.0" | \
  jq '.data.list.vlist[] | {bvid: .bvid, title: .title, play: .play, created: .created, length: .length}'

Parameter Explanation:

```
mid
```
: Uploader's UID
```
ps
```
: Number of items per page (max 50)
```
pn
```
: Page number (starts from 1)
```
order
```
: Sorting method,
```
pubdate
```
(latest) or
```
click
```
(most views)

Batch Pagination Example:

bash

for pn in 1 2 3; do
  curl -s "https://api.bilibili.com/x/space/arc/search?mid=$UID&ps=30&pn=$pn&order=pubdate" \
    -H "User-Agent: Mozilla/5.0" | \
    jq -r '.data.list.vlist[].bvid'
  sleep 1.5
done

7. ASR Fallback Transcription When No Subtitles Are Available (Requires Cookie + SiliconFlow API Key)

When there are no AI subtitles for a video, you can download the audio and call ASR to get the text.

Step 1: Download Audio with yt-dlp

bash

BVID="BV1xx411c7mD"
yt-dlp --add-header "Cookie:$COOKIE" \
  -f bestaudio \
  --extract-audio --audio-format mp3 \
  -o "/tmp/%(id)s.%(ext)s" \
  "https://www.bilibili.com/video/$BVID"

Step 2: Call SiliconFlow ASR for Transcription

bash

AUDIO_FILE="/tmp/${BVID}.mp3"
SILICONFLOW_API_KEY="your_api_key"

curl -s "https://api.siliconflow.cn/v1/audio/transcriptions" \
  -H "Authorization: Bearer $SILICONFLOW_API_KEY" \
  -F "file=@${AUDIO_FILE};type=audio/mpeg" \
  -F "model=FunAudioLLM/SenseVoiceSmall" | \
  jq -r '.text'

Decision Priority: 1.优先尝试获取 B 站 AI 字幕（免费、快） 2.字幕不存在时，才走 ASR 兜底（消耗 API 额度）

8. Optional Data Persistence

The extracted content (subtitles/comments/metadata) is determined by AI whether the user needs to store it. The table structure is maintained by the user, and only a reference template is provided here.

Reference Table Structure

sql

CREATE TABLE IF NOT EXISTS bili_videos (
    bvid        VARCHAR(20) PRIMARY KEY,
    title       TEXT,
    up_name     VARCHAR(255),
    up_mid      BIGINT,
    duration    INTEGER,         -- seconds
    view_count  INTEGER,
    like_count  INTEGER,
    coin_count  INTEGER,
    pub_time    TIMESTAMP,
    content     TEXT,            -- Subtitle or ASR transcribed text
    source      VARCHAR(20),     -- 'subtitle' or 'asr'
    created_at  TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

Storage Example

bash

psql $DATABASE_URL -c "
INSERT INTO bili_videos (bvid, title, up_name, up_mid, pub_time, content, source)
VALUES ('BV1xx411c7mD', 'Video Title', 'Uploader Name', 12345678, '2024-01-01', 'Subtitle text...', 'subtitle')
ON CONFLICT (bvid) DO UPDATE SET content = EXCLUDED.content;
"

Storage Decision Principles:

Only perform writing when the user explicitly says "save", "store in database", or "record it"
By default, only display content and do not automatically write to the database
The database connection string is read from the user's environment variable
```
DATABASE_URL
```

Access Restriction Handling (412 Precondition Failed)

If you encounter a Bilibili 412 error, it means the current environment is under risk control. You need to ask the user to re-obtain the Cookie.

Notes

No video download by default: Only extract information.
Dependency Check: Ensure
```
jq
```
and
```
yt-dlp
```
are installed in the environment.
Cookie Requirements: Only subtitles, barrage, and ASR audio download require a Cookie. Search, video details, and comments do not require a Cookie.
API Limits: The search API has frequency limits. Avoid a large number of requests in a short time; interval of more than 1.5s between pages when batch paginating.
Subtitle Priority: Bilibili automatically generates AI subtitles. Prioritize checking for available subtitles, and use ASR as a fallback only when no subtitles exist.
Storage Decoupling: Persistence is an optional action, not bound to any fixed table structure. Users manage their own databases.

bilibili-video-helper

NPX Install

Tags

SKILL.md Content (Chinese)

Bilibili Video Helper

API Cookie Requirements Overview

Prerequisites: Cookie (Only required for subtitles and barrage)

Core Workflow

1. Video Search (No Cookie Required)

2. Get Video Details (No Cookie Required)

3. Get Comments (No Cookie Required)

Step 1: Get AID

Step 2: Get Comment List

4. Get Subtitles (Cookie Required)

Step 1: Get CID

Step 2: Get Subtitle List

Step 3: Download Subtitle Content (No Cookie Required)

5. Get Barrage (Cookie Required)

Access Restriction Handling (412 Precondition Failed)

6. Get Uploader's Video List (No Cookie Required)

7. ASR Fallback Transcription When No Subtitles Are Available (Requires Cookie + SiliconFlow API Key)

Step 1: Download Audio with yt-dlp

Step 2: Call SiliconFlow ASR for Transcription

8. Optional Data Persistence

Reference Table Structure

Storage Example

Access Restriction Handling (412 Precondition Failed)

Notes