DEV Community

The Twitter account that launched Built on @forem 🌱

On Bluesky @/ DMs — please email us for support!

[DEV Community @ThePracticalDev]

The final Hacktoberfest Open-Source AI Challenge is live! 👻

🎃 Build something spooky, or honor the past for Día de los Muertos
🎃 $2,850 prize pool, 21 winners
🎃 Submissions due Oct. 31

https://t.co/4Cg4AP0qwX

[DEV Community @ThePracticalDev]

@tech_christine How to check your PHP version and what to patch to:

https://t.co/yKiVR6h7RG

[DEV Community @ThePracticalDev]

PHP2Shell (CVE-2026-109618 and CVE-2026-109619) lets an attacker run code on your server without logging in, and the bugs are in PHP itself.

Updating WordPress or your plugins won't fix them. @tech_christine checked which PHP versions are safe. 🧵

[DEV Community @ThePracticalDev]

@tech_christine PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35 fix both bugs. Both are rated 9.2 out of 10 (CVSS v4.0).

Any PHP that serves web requests is affected, including PHP-FPM, Apache mod_php, php-cgi, FrankenPHP and LiteSpeed.

[DEV Community @ThePracticalDev]

One agent in one long chat is the worst reviewer for its own code. @playful_program split their OpenCode setup into a builder and four specialist subagents, each in its own markdown file, and nothing gets edited until they approve it.

https://t.co/v7iSPpa9mG

[DEV Community @ThePracticalDev]

@RandalSchwartz The full write-up, with the flight-recorder data:

https://t.co/9LsHXbchKk

[DEV Community @ThePracticalDev]

A 2026 Harvard study of 718 companies found AI coding agents write 30% more code, but close no extra tickets because PR review times jump 49%.

@RandalSchwartz put his 1986 Commitment Machine into live production across Gemini, Claude and GPT-OSS. 🧵

[DEV Community @ThePracticalDev]

@RandalSchwartz Five live production tickets, a mid-flight vendor swap and an open-weight crash test.

His notes say every code change is paired with an immediate critical reviewer, and that step "really reduces friction."

[DEV Community @ThePracticalDev]

Default Kubernetes settings cut off long AI agent sessions. A 3-minute streaming session dies on 30-second load balancer timeouts and SIGTERM during rollouts. @olivi_eh shows how to fix it on GKE.

https://t.co/JQWnoUpz6f

[DEV Community @ThePracticalDev]

@AlexGeorgiev17 The benchmark and method:

https://t.co/UCw3NPosH1

[DEV Community @ThePracticalDev]

Git 2.56's last-modified command is fast on one directory and slower than the loop it replaces on scattered files.

@AlexGeorgiev17 benchmarked it on git/git: 861 ms vs 2,981 ms for the old per-file loop on one directory, but 6.4 seconds once the same 40 files were scattered. 🧵

[DEV Community @ThePracticalDev]

@AlexGeorgiev17 The command walks history once for all paths, which is the point. It's still marked EXPERIMENTAL in its manual, and a parent-then-child pathspec order silently drops a row.

[DEV Community @ThePracticalDev]

Getting accepted to speak at a tech conference is the easy part.

@SylwiaLask8977 on what comes after: rejected abstracts, months of prep, logistics, and impostor syndrome in a second language. And whether it's worth it.

https://t.co/g91wIOb81K

[DEV Community @ThePracticalDev]

@HarshSh54928171 The Jevons Paradox of AI coding, and how to move the bottleneck back to design:

https://t.co/fhDNJSym9p

[DEV Community @ThePracticalDev]

AI can write an auth flow, schema and tests in 30 seconds.

So why did @HarshSh54928171's last "simple" feature still take three weeks to reach production? 🧵

[DEV Community @ThePracticalDev]

@HarshSh54928171 Writing the code was always the cheapest part. The cost just moved to three places we rarely measure:

1. Review: reverse-engineering AI code before you can trust it
2. Context: how it breaks the other 50k lines
3. Maintenance: code that's correct but weird

[DEV Community @ThePracticalDev]

Getting WebGPU to run on Linux with hybrid Intel + NVIDIA graphics on Wayland is still a fight.

@sizzlebopzz got it working, forgot how, then sent an AI agent to find their own notes. The script and every flag:

https://t.co/fqggoeOhiF

[DEV Community @ThePracticalDev]

@sanity_io @MLHacks These were so great

[Sanity @sanity_io]

335 builds in 19 days.
The Sanity DEV Challenge winners are in. Agents that actually know things, a shelter for adoptable potholes, and a wine cellar with a memory.
🧵 with @MLHacks and @ThePracticalDev https://t.co/TRWL17wLDs

[DEV Community @ThePracticalDev]

What was the highlight of your week? ✨

https://t.co/25jocl5sVO