Skills · Community project

last30days

last30days is an Agent Skill that researches recent discussion across Reddit, Hacker News, GitHub, Polymarket, X, YouTube, and other sources, then organizes the last 30 days by engagement signals.

59.2k StarsMITPythonUpdated 1 day ago
01

Project overview

last30days is an Agent Skill for researching recent discussion. Given a person, product, or topic, it can collect public signals from Reddit, Hacker News, GitHub, Polymarket, X, YouTube, and other sources, then organize the last 30 days into an evidence-aware brief. Core Reddit, Hacker News, Polymarket, and GitHub paths can work immediately; broader coverage requires optional configuration. This page is based on public repository material and does not present project claims as hands-on testing.

last30days README showing research sources such as Reddit, X, YouTube, Hacker News, and GitHub
The README source table explains the signals supplied by each platform and distinguishes core sources from optional integrations.View repository image
02

Core capabilities

01

Last-30-days multi-source research

Queries community, developer, prediction-market, and web sources for a person, product, or topic within a recent time window.

02

Engagement-based ranking

Uses public votes, comments, views, GitHub activity, and Polymarket probabilities to help prioritize discussion.

03

Topic comparison and trend discovery

Supports comparisons between tools and a discovery mode for topics that are gaining recent attention.

04

Source diagnostics and research briefs

Provides preflight, doctor, and source-status reporting so users can distinguish success, partial coverage, missing configuration, and failure.

03

Installation and usage

Install globally through the documented Skills CLI on Codex, Cursor, Copilot, Gemini CLI, and similar hosts. Claude Code can use the marketplace instead. Run the safe preflight first, then verify with a nonsensitive research topic.

AI AGENT INSTALL

Let an AI Agent install it

Send this prompt to Codex, Claude Code, or another AI agent that can work with your local environment.

Help me install the last30days Skill from https://github.com/mvanhorn/last30days-skill. Read README.md, skills/last30days/SKILL.md, CONFIGURATION.md, and LICENSE first. Confirm that npx is available, then run the documented `npx skills add mvanhorn/last30days-skill -g` command for the current agent host. Ask before replacing an existing Skill, installing additional software, reading browser cookies, creating configuration files, storing credentials, or overwriting files. Run the bundled `scripts/last30days.py --preflight` first without reading cookies or making paid requests. Then use the nonsensitive topic “AI video tools” for one minimal `/last30days` research run and report available sources, missing sources, output locations, and verification results.
01Before you start
  • A Node.js environment that can run npx
  • An AI agent that supports Agent Skills
  • Network access to GitHub and at least one public research source
  • Prepare API keys or accounts only when you explicitly enable advanced sources
02Copy the install command or configuration
npx skills add mvanhorn/last30days-skill -g
03Complete the setup steps
  1. 1
    Install the Skill

    Run `npx skills add mvanhorn/last30days-skill -g` on an Agent Skills host. Claude Code users may use the marketplace path, but should keep only one installation method per machine.

  2. 2
    Review permissions and sources

    Ask the agent to run `scripts/last30days.py --preflight` from the loaded Skill directory and review sources, cache destinations, and any planned browser-session access.

  3. 3
    Run the first recent-topic brief

    Enter `/last30days AI video tools` or another nonsensitive public topic. Start with core sources and confirm that the brief includes its research window and source coverage status.

  4. 4
    Enable advanced sources only when needed

    Read CONFIGURATION.md before adding X, YouTube, TikTok, or paid search access. Configure each account, cookie path, or API key deliberately, then rerun doctor.

How to verify the setup

After installation, ask the agent to run the bundled `scripts/last30days.py --preflight` and review configuration sources, planned writes, and available sources. Then run `/last30days AI video tools`; a recent brief with source coverage status completes the minimal check.

Before using it
  • Do not keep both the marketplace plugin and npx Skills copy active on the same host
  • Advanced sources such as X, TikTok, Instagram, and Perplexity can require an account, cookies, API keys, or cost
  • Preflight reports planned permissions and write destinations without reading browser cookies or running live research
last30days README showing installation paths for Claude Code, Codex, Cursor, Gemini CLI, Claude.ai, and Desktop
The repository documents installation and update paths for the Claude Code plugin, Skills CLI, Claude.ai, Claude Desktop, and OpenClaw.View repository image
04

Use cases

SCENARIO 01

Track AI topics

Turn recent Reddit, Hacker News, GitHub, and web discussion beyond model training data into an editorial brief.

SCENARIO 02

Prepare for a meeting

Review a company, product, or person's recent public releases, community reactions, and disagreements.

SCENARIO 03

Compare tools

Place recent community voices, release activity, and public adoption signals for several candidate tools into one report.

05

Assessment

Based on the README, runtime Skill, configuration guide, and releases, last30days turns scattered recent community signals into a repeatable agent research workflow while exposing per-source coverage. Its keyless core sources make it easier to start than tools that depend entirely on paid APIs, but full coverage still depends on accounts, cookies, network conditions, and third-party endpoints. This assessment is based on public material without installation or a live research run.

Why it may be useful

  • Core Reddit, Hacker News, Polymarket, and GitHub paths can start without API keys
  • Documents Skills CLI, Claude Code marketplace, Claude.ai, and Desktop installation paths
  • Preflight, doctor, and source-status reporting make degraded coverage visible

What to know first

  • Installation and a real research task were not executed for this catalog entry
  • Coverage depends on network access, public endpoints, account state, and optional credentials
  • Engagement measures attention, not factual accuracy or an unbiased sample
06

README

last30days


Overview

last30days is an Agent Skill that researches recent discussion across Reddit, Hacker News, GitHub, Polymarket, X, YouTube, and other sources, then organizes the last 30 days by engagement signals. AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary

Getting started

  • Install globally through the documented Skills CLI on Codex, Cursor, Copilot, Gemini CLI, and similar hosts. Claude Code can use the marketplace instead. Run the safe preflight first, then verify with a nonsensitive research topic.
  • Project focus: help an agent research the last 30 days of discussion and prioritize signals such as votes, comments, views, GitHub activity, or market probabilities.
  • Core sources: the README says Reddit, Hacker News, Polymarket, and GitHub can work without API keys.
  • After installation, ask the agent to run the bundled `scripts/last30days.py --preflight` and review configuration sources, planned writes, and available sources. Then run `/last30days AI video tools`; a recent brief with source coverage status completes the minimal check.

Configuration

npx skills add mvanhorn/last30days-skill -g
Read the complete README on GitHub