code: https://t.co/hh8G2XuuhV abs: https://t.co/6x6dYufTVD

kiss kiss kiss…. AH!!! THEY CAUGHT US!! DISTRACTION TIME!!! https://t.co/jec0TV2cP7

@MaziyarPanahi @capetorch @willccbb https://t.co/lqEtV4Votm

🇬🇧 London, let’s build! Fully Connected lands Nov 4–5 at Convene Sancroft, St. Paul’s. Two days on agentic AI to turn prototypes into production with hands-on workshops, live demos, and real case studies. Builders, founders, infra leads. This one’s for you. Tickets below! https://t.co/dyv4QZDmUY

Day 1 workshops. Day 2 speakers and live demos. Venue: Convene Sancroft, St. Paul’s. Seats are limited. Details and tickets here.👇https://t.co/5tYiHuuuHx

GenAI works when you close the loop. (iterate → evaluate → deploy → monitor) In Ep 03 of @CIOonline’s “Art of the Possible,” our own Kimberly Madia joins @BBGIndustry and @awscloud to share the playbook to turn user feedback into retraining and track cost/latency/quality. https://t.co/N5VnLY2pVz

@CIOonline @BBGIndustry @awscloud Listen here: https://t.co/7c0JF8Ith4

POV: You stopped by the @weights_biases popup Code Cafe on August 17 ☕ https://t.co/NdyKRVzSso

@johannes_hage https://t.co/WN23YS4rBZ

Check out the full changelog here: https://t.co/dWQHphNpfQ

Can you spot all 10 differences? Let us know what you find in the replies👇 https://t.co/YjcfLah8lN

And don't forget to grab $400 off your #GitHubUniverse ticket 🎟️ https://t.co/i4s22kk1Qr

The #ForTheLoveOfCode entries are rolling in… and they’re everything we hoped for. 🎶 A terminal jukebox with dancing notes 😂 A Git hook that plays sitcom laugh tracks 👇 Links in the comments https://t.co/6Rtm5goQyz

🎵 Jukebox CLI by Federico Carollo A jukebox built with Rust + Ratatui. Animated notes, colorful UI, full playback control - right in your terminal #ForTheLoveOfCode https://t.co/oRRtJD5ifn

🤣 Git Laugh Track by Sarthak Sanay Add a little lol to your Git workflow and plays a random sitcom laugh track every time you commit #ForTheLoveOfCode https://t.co/u8bn3tWoa0

What are you building? https://t.co/YGYqyvqt64

Ready to onboard your AI peer programmer? 🤖 It's true: GitHub Copilot coding agent works best when you set it up for success—just like a new teammate. Here's how to use custom instructions, MCP, and more to get the best results. 👇 https://t.co/EF1rTkkrwT

The problem: Applying patches was painfully slow. 🐢 The solution: Build a tool so good at merging you barely need patches. 🤝 The result: Git. 🏆 Hear more stories from Linus Torvalds in our interview for Git’s 20th anniversary. ⬇️ https://t.co/qm3ybASwva https://t.co/WfrHl17NDC

Still confused about MCP? 👀 Here's some context. 😉 https://t.co/yok5eabYmg

What is MCP and why is everyone talking about it? In our latest episode of the GitHub Podcast, the team breaks down how the Model Context Protocol is making AI tools more powerful, open, and accessible. Listen wherever you catch your podcasts or https://t.co/MT9PxIJ0r3 https://t.co/2ax5EJbvWX

What's new from GitHub? 👀 Get a first look at our latest product innovations! Join us live on Aug. 21 to see how your team can build, collaborate, and ship better software, faster. 🚀 Register now for a virtual session in your time zone.👇 https://t.co/81ny0PBDtW https://t.co/se8doI5V8O

You've met GPT-5. Now say hello to GPT‑5 mini. 👋 It's rolling out to all plans of GitHub Copilot, starting with Copilot Enterprise and Pro+, others to follow as capacity comes online. Get faster results with less lag time. 🚀 Try it out in @code, GitHub Copilot Chat, and GitHub Mobile. https://t.co/z9epCdXpRR

✓ You’ve got a browser. ✓ You’ve got JavaScript. ✓ You’ve got 13KB. Join the byte-sized battle for glory, swag, and bragging rights. This year’s #js13k theme: black cat 🐈‍⬛ → https://t.co/fA4ZglItUj https://t.co/cVAfpgqjyZ

@HasanAbuKaram Feel free to open an issue on https://t.co/c4Ri2eqOBr or via https://t.co/QQIVUie2kQ. Thanks!

mojo-regex optimizations from Manuel, Apple GPU support updates from Amir, and Q&A with the team! 👀 Catch the August Community Meeting replay if you weren't able to tune in live: https://t.co/PX1NppJpB2

LLM personalities are forged during training. Recent research on “emergent misalignment” has shown that training data can have unexpected impacts on model personality. Can we use persona vectors to stop this from happening? https://t.co/eQ4Wt4ompm

We introduce a method called preventative steering, which involves steering towards a persona vector to prevent the model acquiring that trait. It's counterintuitive, but it’s analogous to a vaccine—to prevent the model from becoming evil, we actually inject it with evil. https://t.co/VJfZ3u7Lrb