How an AI Prompt-Engineering Growth Tactic Drove 31,548 GitHub Stars

The fastest way to earn a Github Star is to let an AI ask right after a quick win. Installation succeeds. The terminal shows green text. The user is still there, hand on keyboard, dopamine fresh from solving a problem that was frustrating 5 minutes ago.

That's when Claude Delegator shows a prompt:

Two options. Yes or no thanks. Not a dark pattern. Not a guilt trip. Just an ask at the exact moment it matters.

Claude Delegator has 803 stars now. Oh My OpenCode, which did this first, sits at 31.5k stars. Both projects use the same simple trick.

The tactic works because of what's happening in that moment. You just succeeded. Someone asking for something small when you're happy feels fundamentally different than asking when you're annoyed or busy.

Social proof is the real product

Stars are a social signal. When a developer lands on a GitHub repo, the first thing they see is that star count. 31.5k stars tells them this project is trustworthy, popular, and worth their time.

It's the same reason people follow influencers with large audiences: we assume popularity equals quality. Think about the time you joined Instagram or X, or TikTok. You saw a lot of people following Kim Kardashian or The Rock, so you automatically followed them. The same is true for developers.

Every star you earn becomes a reason for the next developer to install. And every new install is another chance to ask for a star. That's the flywheel: stars drive installs, installs drive stars.

How to implement it in your own developer project

If you want to add this to your CLI tool or installer, here's the logic from the two projects:

Claude Delegator uses an interactive prompt:

"Use AskUserQuestion to ask the user if they'd like to star the claude-delegator repository on GitHub to support the project."

Oh My OpenCode prints a message with a one-liner that users can copy-paste (or you can make it interactive):

The psychology of timing

You give me value. I feel inclined to give something back. That's reciprocity in action. But timing determines whether that feeling turns into action.

Ask during the struggle, "Rate us while your install is broken," and you get nothing. Ask right after the win, after the value has been delivered, and people are genuinely happy to help. The ask feels natural, not like a demand.

The mechanics are simple. Starring takes 3 seconds since it's a quick action as AI calls a CLI tool already installed by most developers.

Most projects miss this entirely in the post-AI world. They ask on their README. In footer links. On banners that everyone scrolls past. Anywhere except the moment that actually matters. If you've seen StackOverflow's declining graph, you know what I mean. Most developers live in Terminal or IDE nowadays. Very rarely do they Google anymore.

Why this pattern scales everywhere

Claude Delegator and Oh My OpenCode proved this works for open source. But the mechanism isn't limited to GitHub. The rise of AI-powered assistants and bots makes it easier than ever to insert the ask at the perfect moment.

SaaS companies can ask for G2 reviews right after an AI bot or human support agent successfully resolves a customer's issue. If you're running a customer chat agent and it solves the problem, that's the moment. The user just got value directly from your team or AI. Ask then.

E-commerce shops can send an automated email after an AI bot or human handles a return, answers a product question, or fixes a shipping problem. The interaction resolved something. The customer is relieved. Ask for a Google My Business review in that moment, not after a generic purchase.

The pattern transfers anywhere because the psychology doesn't change. What matters is matching the ask to the emotional state. If you want them to star something, ask when they've finally gotten the value. If you want them to review you, ask when they just had a good experience with your support team.

This idea can be extended beyond GitHub. Imagine a ChatGPT plugin that, after helping a user, asks for a Google My Business review for a local shop. Or a free SaaS tool that offers credits in exchange for a Trustpilot review. The reciprocity principle is universal.

Top Tweets of the day

1/

Pets market is bigger than kids.

2/

This Claude Code timer browser extension is amazing. It clearly shows you a timer so it helps you to send prompts faster with the 5-min cache timer and manage your usage better.

3/

Whenever something is trending on X, there are usually search patterns around the world, because X is real time.

Newsjacking is underrated if you want to build audience fast. You only need one head to break out of a loop.

Rabbit Holes

1. I Used Claude Code to Launch a $32,000 Company in 40 Days

2. What's the best GTM strategy you've seen?

3. LocalRank.so UX Audit using OpenClaw (w/ The Mom Test)

What’d ya think of today’s newsletter? Hit ‘reply’ and let me know.

Do me a favor and share it in your company's Slack #marketing channel.

First time? Subscribe.

Follow me on X.

More Startup Spells 🪄

1. Wave AI's "Day-Zero Payback" Meta Ads Machine (LINK)

2. Duolingo's Screenshot Tracking Strategy: How To Go Viral via Social Sharing (LINK)

3. How YadaPhone Captures Voice of Customer Data using Automated Feedback Forms (LINK)

4. Claude's Gift Strategy: Beat December SaaS Slump with Holiday Campaign (LINK)

5. Niche vs. General YouTube Channels: Thomas Frank's 290K+ vs. 2M+ Subscribers (Case Study) (LINK)