GitHub Copilot code review switches from Lite to Balanced today
From 28 September the default effort level for Copilot code review moves to a higher-reasoning model, unless you already pinned your repo or org to Lite.
Tagged
From 28 September the default effort level for Copilot code review moves to a higher-reasoning model, unless you already pinned your repo or org to Lite.
A regex filter meant to block dangerous SQLite commands in Adminer missed statements prefixed with a UTF-8 BOM, letting an authenticated user write PHP straight to a web-accessible directory. Fixed in 6.1.1.
startSpan, getActiveSpan, recordException and setAttributes let you instrument Workers code without threading span objects through every function call.
A public preview lets you cap what Copilot agents can touch — files, networks and credentials — and adds Dev Container support for running them on remote hosts.
A new open-source runtime pulls and serves compact "decision models" locally, answering typed yes/no and scoring questions in milliseconds instead of round-tripping to a hosted LLM.
Two Actions poisoned in the May Mini Shai-Hulud attack came back online in September with the malicious tags untouched, so tag-pinned workflows ran the payload again.
Baseline turns "which browsers support this" into a yes or a no, and it is now built into MDN, the docs, and a linter you can run in CI.
Bare module specifiers work natively in every current browser. For a site with a handful of dependencies, that removes the build step entirely.