Research · What changed, and what people are saying, before you decide. · a tool
The open-source project last30days-skill, by mvanhorn
“AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary” — the project's own words, on GitHub
Paste this into the AI that runs on your computer. It does the install, checks it works, and tells you what to type first. If anything fails, paste the error back to it.
Install last30days-skill (https://github.com/mvanhorn/last30days-skill) on this computer and get it working for me. I am not a developer; explain each step in one line as you go and never skip one. You are Claude Code, running on my machine. 1. Check what this machine already has (git, Python or Node as the project needs, Docker if the README says so). Tell me anything missing and install it, asking me before anything that needs my password. 2. Install it with `/plugin marketplace add mvanhorn/last30days-skill` in Claude Code, or `npx skills add mvanhorn/last30days-skill -g` on another Agent Skills host. 3. Configure it. If it needs an API key or a login, stop and ask me for it; never guess one and never store it anywhere except where the README says. Tool-specific notes: On Claude Code, install with `/plugin marketplace add mvanhorn/last30days-skill`; other Agent Skills hosts use `npx skills add mvanhorn/last30days-skill -g`. Reddit, Hacker News, Polymarket and GitHub work with zero configuration; a short setup wizard on first run unlocks the rest. Smoke-test by asking about a name or topic you already know well and check the sources it cites are real, recent posts. Run its `--preflight` check (or just ask for a health check) to see exactly which sources are configured and which aren't. YouTube needs `yt-dlp` installed separately (`brew install yt-dlp` on a Mac); X needs either an API key or a logged-in browser session it can read. 4. Run the smallest test the README gives, and show me the output. 5. When it works, tell me what to type first, in one line, for this job: Run it on a real name. If anything fails, show me the exact error and fix it before going on.
Install last30days-skill (https://github.com/mvanhorn/last30days-skill) on this computer and get it working for me. I am not a developer; explain each step in one line as you go and never skip one. You are Kimi, running on my machine. 1. Check what this machine already has (git, Python or Node as the project needs, Docker if the README says so). Tell me anything missing and install it, asking me before anything that needs my password. 2. Install it with `/plugin marketplace add mvanhorn/last30days-skill` in Claude Code, or `npx skills add mvanhorn/last30days-skill -g` on another Agent Skills host. 3. Configure it. If it needs an API key or a login, stop and ask me for it; never guess one and never store it anywhere except where the README says. Tool-specific notes: On Claude Code, install with `/plugin marketplace add mvanhorn/last30days-skill`; other Agent Skills hosts use `npx skills add mvanhorn/last30days-skill -g`. Reddit, Hacker News, Polymarket and GitHub work with zero configuration; a short setup wizard on first run unlocks the rest. Smoke-test by asking about a name or topic you already know well and check the sources it cites are real, recent posts. Run its `--preflight` check (or just ask for a health check) to see exactly which sources are configured and which aren't. YouTube needs `yt-dlp` installed separately (`brew install yt-dlp` on a Mac); X needs either an API key or a logged-in browser session it can read. 4. Run the smallest test the README gives, and show me the output. 5. When it works, tell me what to type first, in one line, for this job: Run it on a real name. If anything fails, show me the exact error and fix it before going on.
Install last30days-skill (https://github.com/mvanhorn/last30days-skill) on this computer and get it working for me. I am not a developer; explain each step in one line as you go and never skip one. You are Gemini CLI, running on my machine. 1. Check what this machine already has (git, Python or Node as the project needs, Docker if the README says so). Tell me anything missing and install it, asking me before anything that needs my password. 2. Install it with `/plugin marketplace add mvanhorn/last30days-skill` in Claude Code, or `npx skills add mvanhorn/last30days-skill -g` on another Agent Skills host. 3. Configure it. If it needs an API key or a login, stop and ask me for it; never guess one and never store it anywhere except where the README says. Tool-specific notes: On Claude Code, install with `/plugin marketplace add mvanhorn/last30days-skill`; other Agent Skills hosts use `npx skills add mvanhorn/last30days-skill -g`. Reddit, Hacker News, Polymarket and GitHub work with zero configuration; a short setup wizard on first run unlocks the rest. Smoke-test by asking about a name or topic you already know well and check the sources it cites are real, recent posts. Run its `--preflight` check (or just ask for a health check) to see exactly which sources are configured and which aren't. YouTube needs `yt-dlp` installed separately (`brew install yt-dlp` on a Mac); X needs either an API key or a logged-in browser session it can read. 4. Run the smallest test the README gives, and show me the output. 5. When it works, tell me what to type first, in one line, for this job: Run it on a real name. If anything fails, show me the exact error and fix it before going on.
Install last30days-skill (https://github.com/mvanhorn/last30days-skill) on this computer and get it working for me. I am not a developer; explain each step in one line as you go and never skip one. You are Codex, running on my machine. 1. Check what this machine already has (git, Python or Node as the project needs, Docker if the README says so). Tell me anything missing and install it, asking me before anything that needs my password. 2. Install it with `/plugin marketplace add mvanhorn/last30days-skill` in Claude Code, or `npx skills add mvanhorn/last30days-skill -g` on another Agent Skills host. 3. Configure it. If it needs an API key or a login, stop and ask me for it; never guess one and never store it anywhere except where the README says. Tool-specific notes: On Claude Code, install with `/plugin marketplace add mvanhorn/last30days-skill`; other Agent Skills hosts use `npx skills add mvanhorn/last30days-skill -g`. Reddit, Hacker News, Polymarket and GitHub work with zero configuration; a short setup wizard on first run unlocks the rest. Smoke-test by asking about a name or topic you already know well and check the sources it cites are real, recent posts. Run its `--preflight` check (or just ask for a health check) to see exactly which sources are configured and which aren't. YouTube needs `yt-dlp` installed separately (`brew install yt-dlp` on a Mac); X needs either an API key or a logged-in browser session it can read. 4. Run the smallest test the README gives, and show me the output. 5. When it works, tell me what to type first, in one line, for this job: Run it on a real name. If anything fails, show me the exact error and fix it before going on.
Ask it about a specific person or company you're about to meet and check every claim it surfaces links back to an actual post, not a summary with no source.
Ask it for a health check — it runs every source and tells you exactly which ones are thin and why, instead of you guessing.
Run the same name through a normal search engine first, then this tool, and notice what it surfaced that ranked-by-relevance search missed — and what search caught that it didn't.
mvanhorn/last30days-skill on GitHub, under the MIT licence. 61,631 stars, checked 9 September 2026. Last change 8 September 2026. We did not write it; we checked it, and wrote this page so you can use it.
Once a week, what changed with your competitors and in your niche. One page, every line sourced.
Give it a web address, or a list of them, and it hands back clean, plain-text versions of those pages — headings and paragraphs stripped of surrounding code — so your AI can read them.
Give it a web address and it hands back the page as clean data — plain-text markdown, or JSON shaped the way you ask for — so your AI can read it without wading through a page's raw code.
You give it a question and it goes and researches it properly — pulling from many web sources at once, reading past what any single article says, and writing up a full report with citations you can check yourself.
If you're launching or growing a brand across Asia and the West, a call is where we work out whether it's a fit.