| name | bilibili-hub |
| description | A skill for reading and writing Bilibili data with Python + UV, using bilibili-api-python + aiohttp. It automatically obtains cookies through `browser_use get_cookies` for authentication, so no manual copying is required. Supports video details, subtitles, AI summaries, comments, user profiles, search, popular videos/rankings, following dynamics feed, favorites, Watch Later, viewing history, interactions (likes, coins, triple action), publishing and deleting dynamics, and more. When a user mentions "Bilibili," "bilibili," "Bilibili videos," "Bilibili popular," "bilibili-hub," "get Bilibili data," "Bilibili subtitles," "Bilibili comments," "Bilibili favorites," "Bilibili dynamics," or any scenario that requires programmatically reading or writing Bilibili content, this skill must be triggered.
|
bilibili-hub
Modified from: jackwener/bilibili-cli (Apache-2.0)
This skill simplifies and modifies the original repository as follows:
- Removed the
browser-cookie3 / click / rich / PyYAML / qrcode dependencies
- Changed cookie authentication to accept a
dict directly or read from environment variables, without automatic browser extraction
- Removed the CLI layer (
commands/), QR login, formatters, and related components
- Kept all API methods and wrapped them uniformly as synchronous interfaces (
asyncio.run)
- Core dependency:
bilibili-api-python, a third-party SDK that reverse engineers the Bilibili API
- In the Minis environment, cookies are obtained automatically through
browser_use get_cookies
File Structure
/var/minis/skills/bilibili-hub/
├── SKILL.md
├── pyproject.toml # bilibili-api-python + aiohttp
└── scripts/
├── __init__.py
├── exceptions.py # 6 structured exception types
├── payloads.py # Data structure normalization (normalize_* functions)
└── client.py # BiliClient core class (all API methods)
Authentication Methods
The Bilibili Web API uses three key cookies:
| Cookie | Description |
|---|
SESSDATA | User session (required for read operations) |
bili_jct | CSRF token (required for write operations: likes, coins, posting dynamics, etc.) |
DedeUserID | User ID (recommended) |
buvid3 | Device ID (recommended; reduces the likelihood of triggering risk control) |
Method 1: Automatically retrieve via browser_use (preferred in Minis environments)
- Use
browser_use navigate to open https://www.bilibili.com and confirm that you are logged in.
- Use
browser_use get_cookies to retrieve cookies (raw values do not appear in the conversation).
- Load the offload env file:
. /var/minis/offloads/env_cookies_www_bilibili_com_xxx.sh
export BILI_SESSDATA="$COOKIE_SESSDATA"
export BILI_JCT="$COOKIE_BILI_JCT"
export BILI_USERID="$COOKIE_DEDEUSERID"
export BILI_BUVID3="$COOKIE_BUVID3"
Note: get_cookies only applies to the current page's domain. Navigate to https://www.bilibili.com before calling it.
Method 2: Manually retrieve from browser DevTools
- Log in to Bilibili, then open DevTools -> Application -> Cookies ->
https://www.bilibili.com.
- Find the values of
SESSDATA, bili_jct, and DedeUserID.
- Store them in the Minis environment variables:
BILI_SESSDATA / BILI_JCT / BILI_USERID.
Ways to pass cookies (three methods)
client = BiliClient.from_env()
client = BiliClient({
"SESSDATA": os.environ["BILI_SESSDATA"],
"bili_jct": os.environ["BILI_JCT"],
"DedeUserID": os.environ["BILI_USERID"],
})
client = BiliClient({"SESSDATA": os.environ["BILI_SESSDATA"]})
Quick Start
Environment setup
cd /var/minis/skills/bilibili-hub
uv sync
Calling as a Python library
import os, json, sys
sys.path.insert(0, "/var/minis/skills/bilibili-hub")
from scripts.client import BiliClient
client = BiliClient.from_env()
me = client.whoami()
print("User:", me.get("name"), "UID:", me.get("mid"))
videos = client.search_videos("Python Tutorial", count=5)
for v in videos:
print(f" {v['bvid']} {v['title']} ({v['duration']})")
detail = client.get_video("BV1xx411c7mD", subtitle=True)
print(detail["video"]["title"])
print(detail["subtitle"]["text"][:200])
hot = client.get_hot(count=10)
for v in hot:
print(f" {v['bvid']} {v['title']} 👁{v['stats']['view']}")
API Method Quick Reference
Account
| Method | Description |
|---|
whoami() | Get information about the currently logged-in user |
Video
| Method | Description |
|---|
get_video(bvid, *, subtitle, subtitle_timeline, ai_summary, comments, related) | Get video details (optional subtitles/AI summary/comments/related videos) |
bvid supports a BV number or full URL and is extracted automatically.
Users
| Method | Description |
|---|
get_user(uid) | Get user profile information + following/follower counts |
get_user_videos(uid, count=20) | Get videos posted by a user |
Search
| Method | Description |
|---|
search_videos(keyword, page=1, count=20) | Search videos |
search_users(keyword, page=1) | Search users |
Discover
| Method | Description |
|---|
get_hot(page=1, count=20) | Site-wide popular videos |
get_rank(day=3, count=50) | Site-wide rankings (day: 1/3/7) |
get_feed(offset=0) | Following dynamics feed (login required) |
get_my_dynamics(offset=0) | Dynamics I published (login required) |
post_dynamic(text) | Post a text dynamic (login + bili_jct required) |
delete_dynamic(dynamic_id) | Delete a dynamic (login + bili_jct required) |
Favorites / History
| Method | Description |
|---|
get_favorites() | Get the favorites folder list (login required) |
get_favorites(folder_id) | Get videos in a favorites folder |
get_following(page=1) | Get the following list (login required) |
get_watch_later() | Get the Watch Later list (login required) |
get_history() | Get viewing history (login required) |
Download
| Method | Description |
|---|
download_video(bvid, output_dir, filename=None) | Download the full video (mp4), automatically handling DASH merging |
download_audio(bvid, output_dir, filename=None) | Download only the audio stream (m4a), suitable for ASR transcription |
Download process:
- DASH streams (common): download the video stream and audio stream separately -> merge with
ffmpeg copy -> if merging fails, keep the silent video
- FLV/MP4 streams (rare): download directly, no merging required
- Without logging in, downloads are limited to 480P; after logging in, 1080P is available (premium members can download higher quality)
| Method | Description |
|---|
like(bvid) / like(bvid, undo=True) | Like / unlike (requires bili_jct) |
coin(bvid, num=1) | Give 1 or 2 coins (requires bili_jct) |
triple(bvid) | Perform the one-click triple action (requires bili_jct) |
unfollow(uid) | Unfollow a user (requires bili_jct) |
Error Handling
from scripts.exceptions import (
AuthenticationError,
RateLimitError,
NotFoundError,
NetworkError,
InvalidBvidError,
BiliError,
)
try:
detail = client.get_video("BV1xx411c7mD")
except AuthenticationError:
print("Cookie has expired. Please retrieve it again.")
except RateLimitError:
print("Risk control triggered. Try again later.")
except NotFoundError:
print("Video does not exist.")
except BiliError as e:
print(f"API error: {e}")
Important Notes
SESSDATA is the minimum requirement for read operations. Write operations (likes, coins, posting dynamics) also require bili_jct.- Cookies are usually valid for several days to several weeks. After they expire, retrieve them again through
browser_use get_cookies.
- Bilibili applies risk control to high-frequency requests (HTTP 412). An operation interval of at least 1 second is recommended.
bilibili-api-python is a community-maintained reverse-engineering project, and its interfaces may break when Bilibili updates.- Write operations (coins, triple action, etc.) cannot be reversed. Use them with caution.
