Bilibili Research with the media-crawler-bilibili Skill

Where this fits

Imagine you are a creator strategist preparing a Bilibili content plan. You need to know which formats are getting traction in your niche, which creators are responding to comment sentiment, and whether a topic is saturated. Manually opening search pages, extracting BVs, and copying stats into a spreadsheet is slow. This skill encodes that workflow as agent instructions, so the agent can collect evidence from Bilibili in a structured, controlled way.

Why agents benefit

• Agent can run Bilibili searches with a defined date window and conservative daily sampling, avoiding aimless scraping and overload.
• It can switch from search mode to exact BV detail or creator mode by putting BVs/URLs in BILI_SPECIFIED_ID_LIST and UIDs in BILI_CREATOR_ID_LIST.
• It can choose whether to collect comments, dynamics, contacts, and media separately, recording which limits were enabled.
• It can shortlist by relevance, format, recency, creator fit, and engagement instead of raw views.
• It can disclose when requested video quality BILI_QN may be unavailable due to account access.

Practical scenarios

Topic and format validation

An agent searches a keyword with --platform bili --type search, restricting to a 30-day window. It returns a shortlist of BVs with titles, formats, creators, and engagement. You compare formats and identify which packaging is growing.

Creator benchmarking

For a list of Bilibili Space URLs, the agent enables CREATOR_MODE and collects creator metadata and recent dynamics. This gives a snapshot of posting cadence and audience interaction without manual profile review.

Audience comment mining

Given a competitor's exact BV number, the agent pulls comments and sorts them by themes. This helps answer what questions the audience still has and feeds title and thumbnail ideation.

Add it to your agent workflow

Install the skill in your agent environment:

npx skills add tsingyuai/growth-lab --skill media-crawler-bilibili

Then include a task such as:

Use the media-crawler-bilibili skill. Search Bilibili for AI study tools from START_DAY to END_DAY with a maximum of 20 notes per day. Shortlist by relevance and engagement, then collect details for the top 5 BVs.

Compare before adopting

Before making this a fixed step, compare it with alternatives. Check whether the skill's search mode still matches MediaCrawler's current CLI options, since scripts drift. Evaluate quality signals: does the skill enforce date ranges and sampling limits, or can it easily over-collect? Consider maintenance freshness—last commit, issue activity—and whether the skill fits your existing agent memory and model destination setup. For a side-by-side view of metadata on OpenAgentSkill, see https://www.openagentskill.com/skills/tsingyuai-media-crawler-bilibili. Also compare against simpler REST API scripts if your need is one-off and narrower.

Why it is worth tracking

This skill has meaningful community momentum through its parent growth-lab repository (1,985 stars) and addresses a recurring need: evidence collection from Bilibili with quality controls. It is worth tracking because it offers a clear migration path from manual research to agent-driven collection, and the explicit time-range, deduplication, and quality disclaimers are signs of a mature skill. Re-evaluate it when MediaCrawler upstream changes, when your research questions expand beyond Bilibili, or when you need stricter interaction with comments and dynamics.