Bilibili Video Monitoring

Problem

UP-masters need a recurring view of video performance: play count, likes, favorites, shares, comments, and high-ranking comment feedback. Manually opening Bilibili, copying metrics, and forwarding notes to a contact is easy to skip. This skill turns that weekly reporting workflow into a script: it targets a fixed UID, fetches the latest post data, builds a structured report, and pushes it through configured channels.

How it works

The skill runs with local Node.js and reads a Bilibili cookie from bibili_cookie.txt. It verifies the cookie using the built-in WBI signing algorithm, then retrieves the latest video list, tracks play, like, favorite, share, and comment counts, and collects the top 5 highly rated comments for the latest 3 videos. The report is written to bibili_report.txt. When invoked as a Skill, it can push through two channels: a WeChat PC automation Skill to a specified contact, and the agently-mail Skill via agently-cli to laohuoji5903@agent.qq.com, with a confirmation step before sending.

Boundaries

This is intended for a Windows local setup; the Node path and report path in the source material are fixed. The cookie is sensitive and expires in roughly 30 days, so it must be refreshed. Push delivery depends on external Skills and login state: WeChat must be logged in, and Agent Mail must have completed agently-cli authorization. It does not log in to Bilibili, scrape arbitrary creators, or replace full content analytics; it mainly supports weekly reporting for a fixed account.