Enter a person's name, company, product, or technology. The skill first identifies relevant X accounts, GitHub repositories, subreddits, and YouTube channels, then searches these platforms in parallel. It scores results based on engagement, relevance, and freshness, finally synthesizing a cited briefing. You can continue asking follow-up questions within the same session. Reddit, HN, Polymarket, and GitHub require no configuration to search; X, YouTube, TikTok, Instagram, etc., require third-party services such as the X API, browser login states, yt-dlp, or ScrapeCreators. Data remains on your local machine by default; if you configure keys for inference services like OpenAI or xAI, the topic and retrieved snippets are sent to that service for reranking.
Why it matters Scores results based on likes, engagement, and prediction market odds, then merges discussions about the same topic across platforms to supplement recent community trends missing from training data.
Stars 68138 (+135 / 24h)
Reasons
- 135 new stars in 24 hours, total stars 68138 (indicating attention only); MIT license
- The README includes specific data processing instructions and lists required keys and costs by source
Usage
- In Claude Code, run /plugin marketplace add mvanhorn/last30days-skill and /plugin install last30days; for other agents, use npx skills add mvanhorn/last30days-skill -g, then enter /last30days followed by the topic
Risks
- Searching X can directly use browser login states; some platforms rely on third-party scraping services, so you must determine compliance with each platform's terms yourself
- The README cites checking a specific person's tweets and videos from the last 30 days as a typical use case; be mindful of privacy when investigating specific individuals