Your dependency list does not contain the code most likely to be attacked
Package manifests describe what you chose. Native decoders, fonts and archive readers arrive underneath that, selected by file content rather than by you.
Contributing writer
Software engineer, BSc First Class Honours in Computing and Information Systems. Writes about the JVM, build tooling and the parts of a release note that actually change behaviour.
Package manifests describe what you chose. Native decoders, fonts and archive readers arrive underneath that, selected by file content rather than by you.
A JJ Con 2026 talk surveys what moved from community config into the tool itself, and where the version-control gap against git finally closed.
A year of supply-chain work on a Windows C toolchain: code signing, builds triggered only by a tag push, immutable artefacts, and a multilib x64 release.
Hacktron's HEIF Heist traces attacks on libheif and libde265 through ImageMagick, libvips and Sharp into applications that never chose to decode HEIC.
Jane Street measured how much a sequence's training weight changes what a model learns, and found the relationship is not monotonic across scales.
A study of the popular fast hashes shows attacker-selected inputs can collide far above the average rate, without the attacker ever learning the seed.
The framework's own compatibility layer is gone because Vite absorbed the problem it existed to solve. Migration touches imports, environment types, server runtimes and middleware.
Plus lower memory and CPU on large downscaled JPEGs, hardware H264 for WebRTC on ARM64 Windows, and fixes for DNS over HTTPS and the built-in VPN.
A Go CLI that reads a diff, lets an agent walk the rest of the repository, and returns line-level comments. Apache-2.0, and already on its twelfth minor release.
TLS 1.3 makes the client guess a key agreement algorithm in the first packet. Cloudflare replaced the guess with a per-origin measurement.