About this skill

Problem It Solves

Bilibili danmaku captures real-time audience reactions, but raw danmaku is noisy, temporally scattered, and often too long to feed directly into an LLM. This skill turns a Bilibili video URL into structured, reviewable danmaku data, helping engineers and content operators identify sentiment bias, high-energy moments, meme discussions, and potential public-opinion risks.

How It Works and Limits

The skill uses public Bilibili APIs: it resolves bvid into video metadata, fetches danmaku by cid, and splits the data into 10 timeline segments. It counts positive keywords, negative keywords, and question-like phrases, then samples representative lines from each segment to keep the prompt manageable. The outputs include a JSON data file, a ready-to-use LLM analysis prompt, and a Markdown report. It is best suited for publicly accessible videos with moderate danmaku volume; if cid lookup fails, the video has no danmaku, or requests hit 412 / 403 limits, retry may be needed. The script prepares data and prompts; final sentiment interpretation still requires an LLM or human review.

Screenshots

Use Cases

After a trending video ships, pull danmaku and compare positive, negative, and question shares to judge whether audience sentiment leans supportive or skeptical.
When reviewing an uploader’s video, split danmaku into 10 timeline segments and locate peak density and high-energy moments for a report.
During public-opinion risk checks, extract negative keywords and representative danmaku samples to isolate moments that may trigger controversy.
When drafting LLM prompts, package video metadata, danmaku statistics, and segmented samples into a ready-to-run analysis prompt file.

Best For

Content operations leads reviewing Bilibili videos, who need danmaku sentiment, topics, and risk segments in a report.
Public-opinion analysts who need fast extraction of negative keywords and high-density danmaku windows from public videos.
LLM workflow engineers who need danmaku data converted into structured JSON and standardized analysis prompts.
Product managers studying user interaction, who need to observe memes, questions, and interaction patterns in danmaku.