William Mendonça

Introducing Dimension. The AI coworker that never sleeps. Set it up in 2 minutes and it gets to work. Morning briefings, meeting prep, email drafts, action items, and more. Get started at dimension.dev

We need a name for performative agent parallelization I propose: “slop theater”

Just two Claude Code instances chatting with each other via the tmux API. And it got pretty wild 😆 (gave them a @ElevenLabs voice just for the fun of it)

What are Claude Subagents? Most people use Claude Code like a single employee handling everything in one conversation. That's leaving a lot on the table. Claude Code has a feature called subagents, and once you understand how they work, you'll rethink how you approach complex tasks entirely. Here's the idea. Your main agent is running a session and context is filling up. Instead of doing exploratory work inline and burning through your context window, it spawns a subagent. The subagent spins up with a fresh context window. It sees only its system prompt and the task. It does the deep work in isolation, then returns a compressed summary back to the main agent. All the exploration stays contained. Your main session stays clean. But that's just the mechanical part. The real power is in configuration. Each subagent in your 𝗮𝗴𝗲𝗻𝘁𝘀/ folder gets its own system prompt, tool access, and model preference. A security auditor restricted to 𝗥𝗲𝗮𝗱, 𝗚𝗿𝗲𝗽, and 𝗚𝗹𝗼𝗯. A code reviewer running on Haiku because read-only inspection doesn't need Opus. A documentation writer that only touches markdown files. You're not just delegating work. You're delegating context. And a subagent that can only read files cannot accidentally write them. That constraint is a feature, not a limitation. Three things subagents are genuinely great for: - Deep research where you only need the conclusions, not the full journey of how the agent got there - Parallel investigation where multiple agents look at different parts of a problem simultaneously - Specialized work that needs a focused system prompt conflicting with your main agent's instructions Without subagents, everything piles into one context window. With them, the messy exploration stays contained and only the answer comes back. Running one agent for everything is first gear. Subagents are how you actually scale. The article below is a complete guide to 𝗖𝗟𝗔𝗨𝗗𝗘.𝗺𝗱, custom commands, skills, agents, and permissions, and how to set them up properly.

Best GitHub repos for Claude code that will 10x your next project: 1. Superpowers github.com/obra/superpowe… 2. Awesome Claude Code github.com/hesreallyhim/a… 3. GSD (Get Shit Done) github.com/gsd-build/get-… 4. Claude Mem github.com/thedotmack/cla… 5. UI UX Pro Max github.com/thedotmack/cla… 6. n8n-MCP github.com/czlonkowski/n8… 7. Obsidian Skills github.com/nextlevelbuild… 8. LightRAG github.com/hkuds/lightrag 9. Everything Claude Code github.com/kepano/obsidia…

source code now available github.com/collaborator-a…

Most Claude Code setups fail before the first prompt. Not because of skill — because there’s no structure. No CLAUDE.md No skills No hooks No agents No workspace memory So Claude keeps guessing. And you keep re-explaining. Power users don’t rely on prompts. They build an environment Claude can think inside. This kind of setup turns Claude from: “help me write this” into “ship this entire feature.” If you're serious about Claude Code, save this. You’ll want it when your workflow starts breaking.

Most people using Claude Code are leaving 80% of its power on the table. Here's the fix (and it's free): A developer friend of ours built a 29-agent open-source plugin that turns Claude Code into a full engineering team. It solves the biggest problem with AI builds... Most AI-generated code makes future work HARDER. This plugin flips that with 5 steps: (broken down into skills) • Brainstorm step: agents scan your repo and propose approaches with tradeoffs documented • Plan step: parallel agents fill in gaps, catch edge cases, output a full spec • Work step: swarm agents build in parallel, test after every change, only commit when green • Review step: multi-agent code review with P1/P2/P3 triage before you ship • Compound step: after every build, 5 agents capture the root cause, the fix, and how to prevent it next time…. saved to a searchable KB inside your repo That last step is how each unit of engineering work makes the subsequent units easier. If you’re not technical (like me), this is an absolute game changer. I'm sharing the GitHub link with anyone who wants it. • Comment "repo" • Follow me (so I can DM) And I'll send it over right away.

🚨 URGENTE: Claude agora pode criar todo o seu currículo e perfil do LinkedIn como um recrutador executivo de $500/hora da Robert Half. De graça. Aqui estão 12 prompts que te garantem chamadas para entrevista em até 7 dias: (Salve isso antes que desapareça)

A Anthropic lançou 8 novas ferramentas em uma semana. Não foi só atualização. Foi mudança de nível. A velocidade da IA está ficando absurda.

PASTE ANY URL AND IT TURNS INTO AN EDITABLE UI. EVERY WEBSITE ON THE INTERNET JUST BECAME A DESIGN REFERENCE.

A STUDENT USED CLAUDE TO BUILD A POLYMARKET BOT THAT TURNED $1,400 INTO $238,000 IN 11 DAYS.

GOOGLE JUST UPDATED STITCH AND ITS HONESTLY CRAZY - AI vibe-based UI design partner - New: infinite canvas, voice input, instant prototypes, better context + DESIGN.md - Live now with Gemini — free to try

The new Google AI vibecoding experience in @GoogleAIStudio is goated ngl

@LLMJunky CLI is live!

Not sure how I feel about this to be honest. On one hand, I will be happy to have GPT Pro in "Codex," but on the other, I've really come to appreciate all the focus and attention they've placed on making a purely swe focused product. Tough call. What do you think?

20 days stuck in review. Now it’s finally live. I built Mastery because self-improvement apps feel like homework. You miss a day → you feel behind → you quit. So I removed that entire feeling. Mastery doesn’t treat your life like a checklist. It treats it like a game. Habits become actions you take. Skills become things you level up. Progress actually looks like progress. You’re not chasing streaks. You’re building a character. And the only rule is simple: show up more than you don’t. That’s it. Mastery is out now.

What's new in Remotion? 💥 0:04 Light Leaks 📢 0:30 Sound Effects ▲ 1:12 Render on Vercel 🤹 1:50 New Skills 🌐 2:14 Web Renderer Update 🕴️ 2:24 Less teething with Agents 📦 3:17 Rspack ↔️ 4:07 Sneak Peek: Visual Mode

🚀 Codex 0.116.0 is out! 🔐 TUI: ChatGPT device-code sign-in support 🔌 Plugin auto-install prompts and remote sync 🪝 New hook to block or modify prompts 🎙 Realtime starts with recent thread context Changelog: github.com/openai/codex/r…

Releasing Limux, a lightweight GPU-accelerated terminal workspace manager for Linux. Powered by Ghostty's rendering engine, with split panes, tabbed workspaces, and a built-in browser. Inspired by cmux, re-built completely in Rust, for a snappy experience. Link to repo 👇