Connect your Claude, Codex or OpenClaw for full transcripts on millions of podcasts. Two week free trial, cancel anytime.
Start free trial →The New Stack Podcast · Hands-On Engineering Podcasts · November 2026
Angie Jones maintains the open source project Goose and says most contributions now come with an AI agent's help. This is how the project responds to a flood of low-quality pull requests.
majority. I would say in the 90th percentile, like vast majority. And we had to make a statement to say, we know you're going to use your agent. In fact, we expect it. We would be silly to say you couldn't work on an AI. It's a genetic AI
foundation. That's right.
And so, yeah, we're fine with them using their agent. But now the question has become: we actually switched to an issue first model. So don't just put up a PR, but make an issue. Let's talk about the problem and the direction that we want to take with the maintainers, collaborate on the design. And then who cares? Your agent, my agent, the implementation is trivial.
Yeah, yeah. What about like how are you managing, like the speed has increased as well for these projects now, right? There's so many more of these contributions, right? Yeah, that's even if you go this issue first approach.
There is. So we've had to use agents ourselves to help us maintain. Steve Yegi wrote a fascinating article. If anybody wants to go on the deep end with this, it's called Vibe Maintainer. We're not quite at Vibe Maintainer level, but he has agents that basically do the triaging. They'll patch up the…
I'm sure. Well, you're not running the full gas town. Not just