THE GUY WHO WON ANTHROPIC'S HACKATHON JUST GAVE AWAY HIS ENTIRE CLAUDE CODE PLAYBOOK FOR FREE. 10 MONTHS OF WORK, ALL PUBLIC Affaan Mustafa won the Anthropic x Forum Ventures hackathon by building a full startup in 8 hours with Claude Code. Then he open-sourced the exact setup that... did it. It's called Everything Claude Code, and it turns Claude from one assistant into an entire engineering team Repo: affaan-m/ecc This isn't a prompt pack. It's a system he refined over 10+ months of daily use shipping real products What's inside: A huge library of skills, dozens of specialized subagents, and ready-made commands, all working together. Each piece does one job. One subagent reviews security against OWASP standards. One optimizes memory so Claude stops forgetting earlier decisions around hour three. One learns from your past sessions and projects so the setup gets smarter the more you use it. Others handle planning, test-driven development, and language-specific code review Instead of one assistant writing code, you get an orchestrated team. A main session delegates to the right specialist when the task calls for it, the way a real dev team splits work The best part: it's not locked to one tool. It runs in Claude Code, Cursor, Codex and OpenCode, across Windows, Mac and Linux. Free, MIT licensed This is the difference between using Claude like a search box and running it like a team that ships. The guy spent 10 months figuring out what actually works so you don't have to Bookmark thisshow more

one hackathon winner's actual playbook beats 10 "how to use AI" courses. people who ship > people who explain

totally agree

This is total BS. He did not build a full working startup in 8 hours. Where was the income or the seed capital? If no direct economic value is created then it is called a demo. To be blunt; since the award was by Anthopic the award is called a marketing stunt.

that's the move right there. building in public, shipping fast, then giving the playbook away so everyone else can ship faster too. that's how you actually move the needle instead of gatekeeping a win.

he saved years for some people with this move what a hero

he is real legend alive

Most will scroll past. Fewer will read the thread. Even fewer will click the repo. A tiny fraction will actually install it, experiment, and ship faster. That last group? Their output compounds. Their careers accelerate. They build the stuff the rest just tweet about. Affaan dropped 10 months of battle-tested Claude engineering (subagents for security, memory, planning, reviews — the full team) for free after winning Anthropic’s hackathon. This is one of those rare “actually changes how you work” drops. Repo: (or everything-claude-code) Don’t just bookmark. Try one workflow today. The gap between readers and doers has never been wider — or more valuable.

@grok if i have Claude Code, what do i input into Claude code for this open source set up? But first, if it is Repo: affaan-m/ecc then what should i expect will happen?

This is total BS. He did not build a full working startup in 8 hours. Where was the income or the seed capital? If no direct economic value is created then it is called a demo. To be blunt; since the award was by Anthopic the award is called a marketing stunt.

The very Last Thing you want to do is to start working with Claude Code using not only somebody’s else work but a system with quote “huge library of skills, dozens of subagents” Which you won’t neither fully understand nor know what they’re doing at any given time. Start small

the memory optimization subagent is the real MVP here, claude forgetting context mid-project is the most frustrating thing

It's cool that he doesn't charge for it

yess

I think he won anthropic API credits though

not sure a hackathon setup translates to real product work, 8 hours optimizes for demo not for the week-3 regression you can't explain. has anyone seen this playbook tested outside the hackathon context?

This isn’t another prompt pack. ECC turns Claude into a real dev team: • Security subagent (OWASP) • Memory agent that doesn’t forget at hour 3 • Planner + TDD + reviewer agents One main Claude delegates like a lead engineer. Refined over 10 months, battle-tested in a hackathon startup build. MIT license. Works in Claude Code, Cursor, etc. Stop yelling at one intern. Run a squad. Repo: Most will bookmark. Smart ones ship.

Installing it means giving it PreToolUse hooks that intercept and log every Bash/Write/Edit you make ("continuous learning" +"governance capture"). The capture appears local, but it's a large attack/telemetry surface — worth auditing scripts/hooks/ yourself before running it

The shift isn't from one AI model to another. It's from using AI for individual tasks to building systems that retain context, knowledge, and workflows. When learning compounds, productivity becomes a byproduct rather than the goal.

Most people drop hype threads or $99 prompt packs. This guy actually shipped real stuff for 10 months, won a hackathon with it, then open-sourced the whole system. The subagent + memory + skill architecture is the direction serious Claude Code users are moving toward anyway. Having one well-maintained public reference instead of everyone reinventing the wheel is genuinely helpful. That said, these “entire engineering team” setups still require significant tweaking and oversight to work reliably on complex, long-running projects. The hackathon demo probably looked a lot cleaner than day-to-day usage.

Building a full-fledged startup in 8 hours is the best proof that the era of "just coding" is over. Now the architecture and the ability to conduct agents decide. Having a ready-made free template for Claude Code or Cursor that does not lose memory and checks security on the fly is pure cheat code. Definitely bookmarked.

thanks for sharing , I was just looking for something interesting for my agents

You’re blessed for dropping this for free @tommydansy please check it out, you might need

Amazing

This dude is firing thousands of people live on air 😭

@threadreaderapp unroll

👀

Why care? Just use gstack!

Turning Claude into an orchestrated team with specialized subagents and memory that prevents forgetting mid-project is exactly how you stop treating it like autocomplete. The setups that actually ship consistently are the ones where that team keeps its shared memory and delegation logic alive across different days and codebases so you don't lose the accumulated context every time you start a new session.

这个用了，确实不错

@grok please summarize

Affaan's move is either genius or a cry for help, either way it's a great catalyst to ta lk about what actually works with Claude Code, like the new opus 4.8 updates.

wow never heard about him

这手册估计比项目本身还值钱了

10 months of daily use shipping real products and he just handed it to the internet for free. The repo is getting cloned thousands of times right now guaranteed

That's some serious work. Mad respect. What can't Claude do these days? The engineering crew's getting an upgrade. Who’s trying this setup out first?

The concept of turning Claude into a full team of office workers completely changes how I view AI workflows.

Open sourcing a full setup like this quietly resets expectations for everyone who ships