Bookmarks

Claude に新たに追加された advisor tool を使用すると、通常のタスクは軽量モデルに任せつつ、必要に応じて高性能モデルに相談することで、性能とコストのバランスを最適化できます。この記事では Claude Code 内で advisor tool を活用する方法について紹介します。

https://aid.connpass.com/event/386203/

Fastest enterprise AI gateway (50x faster than LiteLLM) with adaptive load balancer, cluster mode, guardrails, 1000+ models support & 100 µs overhead at 5k RPS. - maximhq/bifrost

Before cache components, the Precompute pattern encoded request-specific data like auth state into URLs to keep pages static. This post documents how it works, where it shows up in practice, and what cache components mean for it.

For eight years, I’ve wanted a high-quality set of devtools for working with SQLite. Given how important SQLite is to the industry1, I’ve long been puzzled that no one has invested in building a really good developer experience for it2. A couple of weeks ago, after ~250 hours of effort over three months3 on evenings, weekends, and vacation days, I finally released syntaqlite (GitHub), fulfilling this long-held wish. And I believe the main reason this happened was because of AI coding agents4. Of course, there’s no shortage of posts claiming that AI one-shot their project or pushing back and declaring that AI is all slop. I’m going to take a very different approach and, instead, systematically break down my experience building syntaqlite with AI, both where it helped and where it was detrimental. I’ll do this while contextualizing the project and my background so you can independently assess how generalizable this experience was. And whenever I make a claim, I’ll try to back it up with evidence from my project journal, coding transcripts, or commit history5.

David Heinemeier Hansson (DHH) is the creator of Ruby on Rails and Omarchy, co-founder and CTO of 37signals (maker of Basecamp and HEY), and the author of se...

A safe, repeatable process for migrating real-time production software. How we transformed high-risk rewrites into in-place delivery with a 99.995% success rate at launch

A Go (golang) library for parsing and verifying versions and version constraints. - hashicorp/go-version

Recently, during a production incident response, I guessed the root cause of an outage correctly within less than an hour (cool!) and submitted a fix just to rule it out, only to then spend many hours fumbling in the dark because we lacked visibility into version numbers and rollouts… 😞

Is your CLI ready for AI agents? Automated checks to ensure optimum agentic CLI use: don't block agents, don't waste tokens, ensure safe use. - Camil-H/cli-agent-lint

Fuik automatically captures webhook payloads and provides a beautiful UI to process them.

Rails engine that catches and stores webhooks from any provider - Rails-Designer/fuik

Sometimes npm packages are compromised and published with malware. Luckily, there are companies like [Socket], [Snyk], and [Aikido] that detect these compromised packages early. The npm registry usually removes the affected versions within hours. However, there is always a window of time between when the malware is published and when it is detected, during which you could be exposed. Fortunately, there are some things you can do with pnpm to minimize the risks.

The Elegant Bundler for Libraries

npm trusted publishing now supports CircleCI as an OIDC provider, joining GitHub Actions and GitLab CI/CD. Maintainers publishing from CircleCI workflows can now eliminate stored credentials entirely and authenticate directly…

A promise that never resolves is a surprisingly clean way to interrupt an async function.

A universal property-based testing protocol and family of libraries, built on Hypothesis

How a $3.2M production disaster taught us that technical excellence doesn’t always align with...

A parser, formatter, validator, and language server for SQLite SQL. Built on SQLite's own grammar and tokenizer - LalitMaganti/syntaqlite

いわゆるAI疲れは、2024年ごろから言われはじめ、2025年から深刻化しはじめた。AI疲れには、キャッチアップ疲れと、判断疲れの二種類があると言われている。（2024年ころのAI疲れはキャッチアップ疲れの方が比率としては大きかった気がする。） AIコーディングは特にAI疲れを引き起こしやすい。 今のAIは、人間に良く似ているが、全く違う知性である 「判断」が、大量に高速に、そして強度の高く、常に求められるようになった。そして油断すると結果だけを提示されるので、判断するための根拠・材料を集めるコストがバカにならない AI時代は、人間からアウトプットの場を奪う

Adapt your own patterns and processes into a reusable piece of content.

Insights and guidance from our engineering team on how Astral secures its tools.

HTML in Canvas API は WICG で提案されている API で、Canvas 内に直接 HTML を描画できるようにするものです。現在の `` 要素にはリッチテキストや HTML コンテンツを描画する標準的な方法が存在しないという課題があります。この記事では HTML in Canvas の使用方法やユースケースについて説明します。

Claude Managed Agents は Claude を自律的なエージェントとして動作させるためのハーネスとインフラストラクチャーを提供します。長時間かかるタスクや非同期のタスクを実行するために使用するのが想定されています。この記事では実際に Claude Managed Agents を試してみた内容を紹介します。

Orchestrate multiple AI agents, terminals, sketches and notes on one infinite canvas. Built for macOS.

Vera: a programming language designed for LLMs to write - aallan/vera