Summary
Search on one side, your own account on the other.
Search, video details and comments are the read half; the other half likes, coins, favourites, comments and downloads under your own credentials.
What it is
A Python server over the bilibili-api library that drives both halves of Bilibili with your own credentials: searching and reading videos, and the account actions you would otherwise click.
What you get
- search_video searches by keyword with ordering and a result count; get_hot_videos returns the current hot list
- get_video_info returns a video's details, and get_video_comments returns its comments paged and orderable by time
- Account actions on a video: like_video, pay_video_coin, triple_video, and set_video_favorite with unset_video_favorite
- Watch Later handling with add_video_to_toview and delete_video_from_toview
- Favourite lists: create_video_favorite_list, public or private, and delete_video_favorite_list
- send_comment posts a comment, and download_video_best_quality saves a video locally at the best available quality
Requirements
Python with uv, and your own Bilibili credentials in a .env file — the write actions run as you. The project is marked as still in development, so try the write tools on an account you do not mind first.
Setup effort
One command plus a key — uvx bilibili-mcp, then supply credentials