More rules should mean better output. That's the intuition. I spent weeks building a comprehensive CLAUDE.md — 200 lines covering naming conventions, security rules, error handling, architectural patterns, import ordering, type safety requirements, and more. I was proud of it. I'd thought through every scenario. Then I scored the output. 79.0 / 100. My carefully crafted documentation was actively
E aí, gurizada! De uns tempos pra cá, eu percebi um burburinho enorme em torno de uma ferramenta que tem chamado a atenção, e não é por menos: o OpenClaw. Eu, que vivo mergulhado nesse universo de IA e automação, gravei um vídeo recentemente, que está lá no meu canal, assista no YouTube, justamente pra desmistificar essa parada. E hoje, vim aqui no Dev.to pra gente conversar um pouco mais sobre o
Uma skill ruim gera código ruim em escala. Uma skill boa gera código bom em escala. A diferença entre as duas não está na ferramenta, está em como a skill foi construída. Quando uma skill é criada sem contexto suficiente, a IA passa a alucinar sistematicamente: gera código tecnicamente válido, mas semanticamente errado. E faz isso toda vez que a skill é chamada, para todo mundo que a usa. Percebi
The blog you're reading right now was built in a single conversation with Claude Code, Anthropic's CLI, in about 30 minutes. No all-nighter, no purchased template, no WordPress. One working session in the terminal. Here's exactly how it went — real code, real commands, and what almost went wrong. My portfolio web-developpeur.com does its job: showcasing my background and projects. But a static sit
Have you ever looked at code you wrote six months ago and thought: "Who wrote this monster?"? Relax, it happens to all of us. In software engineering, writing code that a machine understands is the easy part. The real challenge is writing code that other humans (including your future self) can understand, maintain, and scale. This is exactly where Software Design Principles come into play. In this
Part 1 of 5 in The New Engineering Contract — what it means to lead engineers when AI is doing more of the coding. SWE-CI tested 18 AI models across 71 consecutive commits. Most broke something on commit 47 they'd already broken on commit 1. That's not an intelligence problem. That's a learning system that isn't learning. A paper made me uncomfortable this month. Not because of what it found about