Feature Flags SUCK! Just kidding, they're actually pretty awesome, but if you use them wrong you could certainly be forgiven for thinking they suck. My talk at #kubecon today covered why people can struggle with feature flags, and what to do about it: https://t.co/OL4RoRDRop

Proactively managing your coding agent's context is probably THE best way to improve the quality of it's work. But there's a lot more to it than firing off a `/compact` every now and then. Read on to learn some neat tricks: https://t.co/XqIR13l6JJ

See if you can spot the prompt injection: https://t.co/6fbODiyHQr Or if you can spot the backdoor being added: https://t.co/9HzWnS9wok More on the lethal trifecta:

Super interesting proof-of-concept exploit against an AI agent via GH pull request. https://t.co/zrmD3n05SN Very sneaky prompt injection which convinces an AI coding agent to subtly inject a remote backdoor into a codebase. A great example of @simonw's Lethal Trifecta

Good example of why it's important to monitor your LLM-powered features in prod. Some bug in Amazon's AI assistant where the product description isn't in it's prompt. Pre-production evals are v. important, but don't help here. Monitoring some key metrics in prod would.

A quick summary of what to "vibe code" without a developer ... and what not to ... 1⃣ If you can buy it, buy it. You cannot build something better than what 10,000+ folks pay often a trivial sum for. And you certainly can't maintain it properly. You can't vibe a better

https://t.co/XQ71kS7V6w

Devs, who is footing the bill for your AI tools? 🧐

“Listen, I’m wondering if our engineers are doing enough with AI…” your CEO says, as the two of you sit down for your 1:1. Wouldn't it be great to have a true strategy in place for adopting AI-assisted coding? I've got you covered! https://t.co/zJBHSzLkqg

بعيد عن هري كتير ... ده أفضل مقال قرأته عن سبب ان فيه ناس شايفه الذكاء الصناعي بيساعدها في شغلها وناس تانية على النقيض شايفاه مش مفيد. كل الشكر لزميلي العزيز اللي خصص ايميل عشان يشارك المقال معي. يستاهل يتبعت في إيميل مش رسالة على WhatsApp أو Slack. https://t.co/jqohcpd619

There's also a REALLY interesting case study from eBay's @OpenFeature adoption that's well worth reading. - Rolling out a platform to 300 teams - Billions of flag evaluations made per day - 2500+ experiments behind flags every year

I got to write the forward for @GetFlagsmith's eBook, Scaling Feature Flags! https://t.co/f6KKK9Lwbp Feature flagging is becoming more and more mainstream, but teams can struggle with some common challenges when adopting the practice. I think this handy little guide will help!

Loads of detailed practical advice in here: https://t.co/1mf7jVGHhL

Engineering leaders at startups sometimes have to fake it as product leaders too 😅 A short guide on what a useful, succinct product strategy looks like, how to build it, and how to use it to identify and prioritize and roadmap of engineering work https://t.co/cEXJ1wUPJT

full-on vibe-coding doesn't work for real-world codebases, but we can still get a lot of the productivity boost via a more iterative approach. Chain-of-vibes lets you scale up agentic AI to non-trivial coding tasks, by keeping an engineer in the loop https://t.co/fSxJveU0re

Why so many teams got a prototype of amazing AI-powered feature built in, like, a week, but then struggled to make improvements after the MVP launched. Fixing this is all about building a systematic feedback loop, using domain-specific evals. https://t.co/6zjkJbJORo

Slides from my Migratory Patterns talk at KubeCon today. I told attendees I'd share the slides - if you weren't at the talk then these probably aren't super helpful on their own. I'll reply in thread with the talk video once it's online. https://t.co/3N6IDkgHCK

LLMs are like a cocky grad student at a bar: confronted with a topic it doesn't know much about, the AI will start to confidently spout convincing bullshit - uh sorry, "hallucinate". Can we fix this by giving the LLM a sense of how certain it feels?

I suspect this design decision won't age well

Feature Flag Super Stream is LIVE right now! Come join us!