The last30days-skill repository is an AI agent-powered research tool that aggregates real-time information from multiple social platforms and data sources to synthesize grounded summaries on any topic. Written in Python, it bridges the gap between walled-garden platforms by bringing together data from Reddit, X, YouTube, TikTok, Instagram Reels, Hacker News, Polymarket, GitHub, Bluesky, Threads, Pinterest, and web search, scoring results by actual user engagement rather than editorial curation.
The core premise addresses a specific problem: traditional search engines like Google rely on editorial signals and stale training data, while individual AI systems have fragmented access to different platforms. Reddit upvotes, X likes, YouTube transcripts, TikTok engagement metrics, and Polymarket odds backed by real money represent millions of people voting with their attention and wallets daily. The skill searches all these sources in parallel and synthesizes them into a single brief, enabling users to understand what's actually happening in real time rather than what search algorithms or editors decide matters.
The tool operates with zero configuration for core sources like Reddit, Hacker News, Polymarket, and GitHub, with a 30-second setup wizard for additional platforms including X, YouTube, and TikTok. It's available across multiple AI agent platforms including Claude Code, Codex, Cursor, Copilot, Gemini CLI, and over 50 other Agent Skills hosts. The skill spec lives in skills/last30days/SKILL.md and represents the source of truth for current command behavior and setup requirements.
Version 3 introduced shareable HTML briefs that generate self-contained, dark-mode, print-friendly files with inline CSS and no JavaScript dependencies, making outputs suitable for Slack, email, or Notion integration. The engine accepts plain language requests and saves briefs to a configurable memory directory, defaulting to ~/Documents/Last30Days/. The HTML output includes badges, inline metadata, full citations, and a colophon noting the topic and re-run instructions.
Real-world use cases span meeting preparation, hiring signal analysis, tool comparisons, and rapid learning. Users run queries like "/last30days Peter Steinberger" to surface recent career moves, GitHub activity, and community discussion across platforms, or "/last30days OpenClaw vs Hermes vs Paperclip" to compare tools with live GitHub star counts and architectural analysis. The tool has been used to understand breaking news, plan trips by aggregating current wait times and community sentiment, and learn emerging techniques through synthesized community consensus.
GitGenius tracking shows the repository has grown from 48,913 to 48,917 stargazers since July 4, 2026, with median issue and PR response latency of 24.4 hours and mean latency of 210.3 hours across 171 tracked items. The most active labels are bug (14 occurrences), enhancement (11), and help wanted (1). Primary contributors include mvanhorn with 121 tracked events, tmchow with 95 events, and DynamycSound with 14 events. The repository overlaps with contributors from koala73/worldmonitor, openclaw/openclaw, and nousresearch/hermes-agent. GitGenius classifies the repository under Alexa, AWS, Billing, Costs, Cloud, Voice, Serverless, Expenses, Tracking, and Reporting categories, though these appear to reflect broader platform categorization rather than the tool's primary function.
