GitHub Killed Its Stargazers API. That’s the Best Thing That Could Have Happened.

You woke up one morning and GitHub’s stargazers API was gone. No more at-scale discovery. No more crawling who follows whom. The tool you relied on to find emerging open-source projects—dead.

And honestly? It’s about time.

For years, we’ve been lying to ourselves. We treated GitHub stars like they meant something. Like a repository with 50,000 stars was inherently more trustworthy than one with 500. We built discovery systems on top of a metric that any bot farm could inflate, any hype cycle could manipulate, and any viral tweet could distort.

Vanity metrics were never trust. They were just noise with better branding.

The API policy change feels like a loss because it is one—of convenience. You can’t do large-scale stargazer analysis anymore. But what you gained is something far more valuable: the death of an illusion.

Here’s what’s actually happening: the constraint is forcing a shift from external, scalable discovery to internal, trust-based signals. And that’s not a downgrade. It’s an upgrade.

The Problem With Scale

Let’s be honest about what the old system actually did. You could pull stargazer data at scale, map networks of followers, and identify “emerging” repositories based on who was starring what. It felt sophisticated. It felt data-driven.

But it was also fundamentally lazy.

You were outsourcing your judgment to a crowd. And crowds are wrong as often as they’re right. They star repositories because of README aesthetics, because of Twitter momentum, because the name sounded cool. They unstar nothing. The signal-to-noise ratio was atrocious.

When the platform takes away your crutch, you finally learn to walk.

What Actually Works Now

The team at Agent Swarm did something interesting when the API changed. Instead of complaining, they went internal. They looked at their own data—the repositories that their trusted stargazers were already following—and built a discovery system from that.

The results? A curated list of 25 FOSS repos focused on agentic infrastructure. Not the most-starred projects. Not the trendiest. The ones that people who actually build things actually care about.

This is the model going forward. Community-driven selection. Trusted curation. Reputation systems that bypass centralized metrics entirely.

The repositories that matter aren’t the ones with the most stars. They’re the ones that practitioners in your specific domain actually use, discuss, and depend on.

The New Moat

Here’s the provocative part: this constraint is actually a competitive advantage.

If you’re building tools, writing analysis, or making investment decisions in open source, the old API was available to everyone. Your insights were commoditized. Anyone could pull the same data and reach the same conclusions.

But trust-based signals? Those are inherently harder to scale. They require actual relationships with communities. They require knowing who matters in a specific niche. They require the kind of domain expertise that can’t be scraped from an endpoint.

The best repositories aren’t the most starred. They’re the ones people actually use.

The organizations that figure this out first—who build networks of trusted curators, who understand which communities matter for which domains—will have a discovery advantage that no API can replicate.

Stop Mourning the API

The frustration is real. You lost a tool. But you also lost an excuse.

For too long, we’ve treated platform-provided metrics as a substitute for judgment. We’ve let GitHub’s star count do our thinking for us. We’ve confused popularity with quality, scale with relevance.

The API change is forcing us back to basics: talk to people, build trust, curate with intention.

That’s harder. It’s slower. It doesn’t scale as cleanly.

But it’s also how we find the projects that actually matter.

The stargazers API died. Real discovery just began.