GitHub – mvanhorn/last30days-skill: Claude Code skill that researches any topic across Reddit + X from the last 30 days, then writes copy-paste-ready prompts

If you’ve spent any time experimenting with AI tools, you’ve probably felt this quiet frustration. You find a “great” prompt… and it’s already outdated. The AI world moves fast. Blink, and the advice you saved last month suddenly feels stale.

That’s where this GitHub project quietly shines.

last30days is a Claude Code skill built by mvanhorn that does something refreshingly practical. It researches any topic across Reddit, X, and the wider web from the last 30 days, then turns what people are actually discussing, upvoting, and sharing into copy-paste-ready prompts you can use right now.

Not theory. Not recycled blog posts. Real, current conversations.

I love this approach because it mirrors how we actually learn. We don’t master tools by reading documentation alone. We learn by watching what works for others, noticing patterns, and stealing ideas (politely). This tool does that homework for you.

Say you want better ChatGPT prompts for legal questions. last30days notices the recent panic around hallucinated case law, the fines, the public embarrassment. Then it builds prompts that emphasize grounding, uncertainty flags, and procedural thinking. The result is safer, more useful output that feels like it came from someone who’s been paying attention.

Or maybe you’re researching a product like ClawdBot. Instead of marketing claims, you get real-world use cases, engagement stats, setup advice, and the little gotchas users keep mentioning. The kind of insight you’d normally gather after hours of scrolling.

It even shines in creative workflows. UI mockups. Music prompts. Image generation. The skill learns how communities are prompting today, then applies those techniques to your specific idea. That’s a subtle but powerful shift.

If you’re curious, you can explore the project directly here:
https://github.com/mvanhorn/last30days-skill

Looking ahead, tools like this hint at a calmer future with AI. Less guessing. Less outdated advice. More grounded, human-informed workflows that evolve as fast as the conversations around them. And honestly… that feels like progress we can all use.