Fetching from the wire…
Top 5 · 2026-08-29 · source-backed
Carson Gross tagged htmx v4.0.0 on August 28 after eight months of work, skipping 3.x entirely to keep an old promise never to ship a backwards-incompatible htmx 3. The repo sits at 49,061 stars and the announcement reached 691 points on Hacker News. (htmx)
The breaking change the team itself flags as the biggest upgrade burden: attribute inheritance flips from implicit to explicit. hx-target on a parent no longer applies to children. You write hx-target:inherited="#div" if you want the old behavior. There's a CLI tool that finds the places needing the modifier, and an htmx-2-compat extension that restores the old semantics while you migrate.
Underneath, fetch() replaces XMLHttpRequest as the ajax layer, which takes the entire htmx:xhr:* event family with it and renames events to an htmx:phase:action shape. localStorage history caching is gone. Idiomorph morphing and an <hx-partial> tag are native now. 2.x stays latest on npm until early 2027, so nobody gets force-upgraded.
I like this release for the reason most people will dislike it. Implicit attribute inheritance is the kind of convenience that saves you keystrokes on day one and costs you an afternoon of debugging on day two hundred, when a hx-target four levels up in a partial you didn't write silently changes where your response goes. Making it explicit is correct. Making it explicit in a way that breaks every app is also correct, because there's no version of that fix that doesn't.
The detail I haven't stopped thinking about is the other artifact in the release. htmx 4.0 ships four skills files written for LLMs: general guidance, debugging, extension authoring, and 2.x upgrading. Not documentation that happens to be readable by a model. Files whose stated audience is the agent.
That's a new maintainer obligation appearing in real time, and it appeared twice this week from different directions. JetBrains published go-modern-guidelines, an official skill teaching agents to stop writing outdated Go. And a Claude Code skill encoding Refactoring UI's design rules as bounded scales reached 412 stars in three days. (GitHub) Spacing, type, color and shadow expressed as constrained scales an agent must pick from, with hierarchy carried by weight and color instead of borders. Taste as a constraint file.
That last one lands close to how I actually work. My rule against accent-colored borders isn't a preference I can explain to a model in prose. It's a constraint, and it holds better when it's written as one. If your framework has migration knowledge that lives in a maintainer's head, the shape it needs to take next is a file an agent can read. htmx figured that out before most.
Each link below shares sources, entities, or timing with this story.
Issue 6235 on anthropics/claude-code asks Claude Code to read AGENTS.md, the config file that Codex, Amp, Cursor and most other harnesses already load, rather than only CLAUDE.md. It has been open since August 2025. It has accumulated over 5,200 reactions and 300+ comments, ma...
One Claude Code release fixed two independent permission-check bypasses on the same day. That's the story. Version 2.1.221, shipped August 4, patches a Bash tool bypass where zsh could execute hidden commands embedded inside [[ ]] regex conditionals. The approval prompt never...
A spec is a press release until someone who didn't write it implements it. GitHub made Agent Plugins 1.0 generally available on August 12 across VS Code, Copilot CLI, the Copilot SDK, and the Copilot app on all plans. The spec, published August 6, was co-authored by AWS, Anysp...
43.3% on Frontier-Bench v0.1. Opus 4.8 scored 18.7%. That's not an incremental bump, that's the same benchmark with a different shape of answer. Anthropic released Claude Opus 5 on July 24 at $5/$25 per million input/output tokens, exactly half of Fable 5's $10/$50, while matc...
The repo (1,861 stars, +314 today) ships a SKILL.md covering useful features from Go 1.0 through 1.27, including everything the modernize analyzer targets, so agents use max(a, b) over if-else, slices.Contains over manual loops, and Go 1.26 additions like new(42) and errors.As...
The headline number on this repo is 65% token savings. The number you should actually care about is 33.2%, and the reason to trust the project is that the maintainer tells you the difference. JuliusBrussee/caveman cut v2.1.0 on August 16 at 19:22 UTC (GitHub). The GitHub API r...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.