<< All versions

Skill v1.0.0

currentAutomated scan100/100
tsingyuai/growth-lab/media-crawler-kuaishou
──Details
PublishedSeptember 29, 2026 at 08:16 AM
Content Hashsha256:40834d982f33c1e7...
Git SHA
──Files
Files (1 file, 1.3 KB)
SKILL.md1.3 KBactive
SKILL.md · 16 lines · 1.3 KB

version: "1.0.0" name: media-crawler-kuaishou description: Collect Kuaishou competitive evidence with MediaCrawler using keyword search, exact video detail, comments, media, and creator profiles. Use for trend, format, audience-language, or creator research requiring reproducible source records and explicit collection limits.


Kuaishou collection

Follow media-crawler. Require purpose, keywords/links, target count, time window, comments/media scope, and a destination in the invoking Model's Memory.

Search with --platform ks --type search, starting serially with one intent-distinct keyword. Shortlist for relevance, format diversity, recency and engagement; do not confuse a search excerpt with full detail. Put full kuaishou.com/short-video/... URLs or pure video IDs in KS_SPECIFIED_ID_LIST; put profile URLs or user IDs in KS_CREATOR_ID_LIST. Run --type detail or --type creator accordingly.

Enable comments/media only for shortlisted items. Verify video ID, creator, caption, publish time, engagement, media association and comment coverage. Deduplicate by video ID, preserve query provenance, and report empty/risk-controlled/partial results honestly. Stop on challenge or expired login.

Return raw/copied paths, exact inputs, selection rationale, counts, coverage, collection time and upstream commit.

All versions