Marcio Castilho

I'm excited to announce my new role as CoFounder and Chief Architect of ReadingMinds.ai. Join the Beta Waitlist now!

@aakashgupta I built and released this app without writing a single line also. Over a 1000+ PRs in 45 days, zero human code. chatml.com

Karpathy just mass-distributed a product without writing any code for it. The gist is one page of markdown describing how to build an LLM-maintained personal wiki. Raw docs go into a folder, an LLM "compiles" them into 100+ interlinked articles, and the whole thing self-heals through automated linting passes. His version runs on Obsidian with custom scripts and 400K words of compiled research. He calls it an "idea file." One page of markdown. Your agent reads it and builds the entire system customized to your stack in an afternoon. His Obsidian setup is probably thousands of lines of hacky Python. You'll never see any of it. A one-page spec and a capable agent gets you 90% of the way there, tuned for your own tools and workflow. The economics of this are wild. In the old model, someone shares a repo and maybe 2% of people actually clone it, install the dependencies, and get it running. Idea files flip that ratio. The agent handles all the translation work between "what Karpathy built" and "what works on your machine." The conversion rate from "saw a cool project" to "have my own version running" collapses from days to hours. Karpathy is basically saying the era of "here's my repo, go figure out the dependencies" is being replaced by "here's what I built and why, go tell your agent to make you one." Software distribution is becoming a game of telephone where every recipient gets a better version than the original.

SQLite Concurrency in Go: What We Learned Building a Desktop AI IDE #BuildInPublic #claudecode #sqlite #golang chatml.com/blog/sqlite-co…

ChatML - The Cockpit for Claude Code See everything that your agents build chatml.com #BuildInPublic #OpenSource #AIEngineering #ClaudeCode

We Built ChatML in 60 days. Here is our story. 750+ Pull Requests. Zero lines by a human. #buildinpublic #claudecode #aiengineer #producthunt chatml.com/blog/we-built-…

@chatmlx visibility into what agents are actually doing is such a pain point right now. love the focus on observability — most tooling treats AI as a black box. congrats on the PH launch!

@chatmlx this is the kind of post that makes building in public worth doing. nice work

@chatmlx A cockpit for Claude Code that shows what your agents are building in real time — that visibility layer is exactly what's missing from most AI dev workflows. Is it showing diffs, tool calls, or both?

ChatML is live on Product Hunt! 🚀 The cockpit for Claude Code: → See every agent, every diff, one dashboard → AI code review with inline comments → 100% free & open source Built entirely by AI. 900+ PRs. Zero human code. producthunt.com/products/chatm…

It's completely free. No accounts. No telemetry. Your code never leaves your machine. GPL-3.0 — fully open source, no enterprise paywall, no catch. Try it → chatml.com Star it → github.com/chatml/chatml Hunt it → producthunt.com/products/chatm…

The code review alone is worth it: → Quick review: fast sanity check → Deep review: logic, edge cases, patterns → Security review: vulnerabilities, injection, auth issues All with inline comments on the actual diff. Before it becomes a PR.

Today I'm launching ChatML on @ProductHunt. I was running 10+ Claude Code agents in parallel and had zero visibility into what they were building. So I built the cockpit. Here's the story — and why every line of code was written by AI. 🧵

We Built an Entire Product with AI. 750+ Pull Requests. Zero Lines Written by a Human. chatml.com/blog/we-built-… #claude #ai #builders @claudeai

Today I am releasing ChatML. A new open source AI Agent Orchestrator for Claude Code. Try it out: chatml.com

@devajmeireles @marcelpociot AJ me segue. Queria te mandar algo relacionado ao que o polyscope faz.

Amazing, @marcelpociot! Thanks for Polyscope. I loved it! I'm already testing it. getpolyscope.com

@VTchuiev @DannyAziz97 I have been using git worktrees to have multiple Claude Code instances.

@DannyAziz97 Always wondered... How people run this without the agents tripping over each other and overwriting each other's code?

I saw a guy coding today in his terminal. Panel 1 Claude Code - refactoring auth. Panel 2 Claude Code - new API endpoint. Panel 3 Claude Code - fixing that production bug. Panel 4 Claude Code - database migration. Panel 5 Claude Code - "quick" CSS update. No branches. Straight to main. --dangerously-skip-permissions on repeat. Like a psychopath. It's me.