AI Agent Skill for Global Hot‑Topic Tracking and Data‑Driven Insights

1. Beyond Simple Aggregation – Deep Analysis and Data Perception

The /last30days skill distinguishes itself from basic aggregators by employing a "multi‑signal quality ranking" and a "data‑driven narrative" that go beyond merely collecting links, instead extracting genuine community dynamics.

Core data sources include:

Reddit, X, YouTube, Hacker News, Bluesky – tracking discussions, high‑upvote content, and video viewpoints.

Polymarket prediction markets – integrating real‑money betting odds to obtain event probability data.

Instagram Reels and TikTok – covering short‑video trends via the ScrapeCreators API.

For example, when researching the "Iran situation," the tool returns not only social‑media opinions but also fifteen live prediction‑market odds, linking viewpoints with monetary bets to produce a more solid analysis.

2. Technical Architecture: Intelligent Scoring and Automatic Discovery

The project's highlight is a composite scoring pipeline. Each result from a platform passes through a unified evaluation system comprising:

Bidirectional text similarity with synonym expansion.

Interaction‑heat normalization.

Source authority weighting.

Cross‑platform convergence detection using mixed‑triple‑gram Jaccard similarity.

Time‑decay factor.

For Polymarket entries, five additional weighted factors are applied: text relevance, 24‑hour trading volume, liquidity depth, price‑change speed, and result competitiveness.

"AI's world evolves monthly. This skill keeps you in sync by researching your topic, surfacing what the community truly up‑votes, shares, bets on, and discusses on camera, then writing a grounded, citation‑rich narrative." – Project README

Version 2.9 adds a "smart sub‑community discovery" feature that automatically finds relevant Reddit subreddits and highlights top comments, contributing an extra 10 % scoring weight.

3. Quick Start: Seamless Integration into AI Workflows

Installation is straightforward, especially for Claude Code users:

Claude Code : add the repository in the plugin market and invoke /last30days [your topic] in a conversation.

ClawHub : run a single Bash command to install.

Comparison mode : use /last30days Claude Code vs Cursor to run three parallel researches and receive a side‑by‑side report with pros/cons tables and data‑driven conclusions.

Projects can be configured by placing a .claude/last30days.env file at the repository root to manage API keys. Each run automatically saves a Markdown report to ~/Documents/LastDays/, gradually building a personal research library.

4. Who Benefits? Not Just Developers

Technical decision‑makers and entrepreneurs can quickly gauge community buzz and real‑world feedback for technologies such as "Rust vs Go" to inform selection or investment.

Content creators and market analysts can track emerging topics like "AI video generation," capture controversies and KOL opinions, and efficiently produce deep content or market reports.

Curious learners can demystify cutting‑edge jargon (e.g., "Seedance 2.0" or "Nano Banana Pro") by letting the AI handle the entire information‑gathering‑to‑summary pipeline.

In essence, anyone who needs continuous monitoring of web dynamics, dislikes information overload, and seeks structured insights can treat this tool as an "external brain" that boosts cognitive efficiency.

5. Outlook

As AI tools themselves evolve rapidly, the /last30days project illustrates a clear direction: future AI agents should not remain mere conversational bots or simple task executors; they must become powerful copilots that perceive the complex world, integrate multi‑source information, and provide decision‑support.