Plain-English explainers on new developer tools, releases and research.
A developer's checklist from the August 2026 Baseline digest: three features that are now safe to use, what they replace, and what to drop from your fallbacks.
Claude Code may put your account email in User-Agent headers on web requests. Here is who is exposed, what the built-in mod does and misses, and what to do now.
How to run an eval loop with Agents CLI: set up the scaffold, compare runs, catch regressions nightly, and avoid fooling yourself with small datasets and summary means.
Cloudflare's Q3 2026 report shows short, small attacks surging while 1 Tbps+ floods fell. Here is which alerting, rate-limit and mitigation settings to recheck.
A plain-English look at provable inference, the FlashProve prototype, its proof sizes and GPU costs, and what builders on third-party LLM APIs can do with it today.
Cloudflare now separates personal AI agents from generic bot traffic and adds per-agent access policies. Here is what site owners can check and set this week.
GitHub moved busy repos to Azure by serializing writes per reference, not per repo. What changed in the Spokes quorum, virtual replicas, and what teams should expect.
Mozilla's WAICT prototype aims to let sites bind client-side code to a publicly logged manifest. Here is what it means for developers and what to evaluate first.
Mozilla explains how it turned AI models into a Firefox bug-finding pipeline: the harness, the bug classes found, the triage load, and what other projects can copy.
Firefox 151 on desktop now supports Web Serial. Here's what it enables, how to feature-detect it with a graceful fallback, and what gaps remain for your project.