Kuaishou Content Research

Problem

When researching Kuaishou topics, competitor content, or trends, keywords, hot lists, titles, authors, and engagement signals are often scattered. Manual review can miss traceable evidence. Judging whether a topic is rising, or comparing how different accounts frame similar topics, becomes slow without stable inputs and a clear result structure.

How It Works

The skill is built around two read-only interfaces:

• Hot search: kuaishou hot-search returns the current Kuaishou hot list with rank, heat signal, titles, authors, and content identifiers.
• Search: kuaishou search --keyword retrieves video samples for a topic or keyword, supports --pages, --max-items, and uses next_page_token for continuation.

At runtime it calls the backend via the SOCIALDATAX_API_KEY environment variable, without relying on local browsers, login state, or posting permissions. Results are best organized as visible evidence + judgment: list hot-list signals, related samples, authors, links or photo_id, and engagement data, then mark follow-up questions. For review, keep source fields and avoid presenting inference as fact.

Boundaries

The skill is read-only and does not log in, comment, like, or modify accounts. On insufficient_balance, do not retry repeatedly; follow the returned top-up link and then continue the original command. Search has no --all full-scan boundary, so pagination must return the complete next_page_token unchanged. If results are thin, broaden keywords or add scene, audience, brand, or time constraints.