Claude Code Anonymous 💒 Cologne Edition - Round 3
Opus 4.5 just dropped. 10.6% better on Aider Polyglot. Outscored every human on Anthropic's internal engineering assessment. The bar keeps moving.
But here's the real question: After months of building with agentic coding tools - what actually stuck?
Round 3 theme: Patterns That Stuck
We've had the hype. We've had the experiments. Now we want to know: Which workflows survived daily use? Which got quietly abandoned? What do you do differently now vs. 6 months ago?
This is Claude Code Anonymous - a safe space for honest, unfiltered technical conversations about building with AI coding agents.
Why "safe space"?
In public: polished threads, impressive demos, "10x productivity" claims.
In private: "How do I stop it from hallucinating imports?" and "It just mass-deleted my tests."
We're here for the private conversations. No recordings. No judgement.
Previous rounds
Round 1 & 2 featured lightning talks on real-world agent debugging, prompt engineering patterns, when to trust (and not trust) agent output, and honest productivity assessments. Discussions ran until 10pm both times.
👉 Browse previous presentations
What makes this different
Nothing is recorded or live-streamed
What's said in the room, stays in the room
Only builders with real hands-on experience
Lightning talks from participants (5 min, 1-2 slides max)
Open discussions about what actually works (and what doesn't)
Format
18:00 - Arrive, get-together, kefir & pizza 🥛🍕
18:45 - Intro and lightning talks
19:45 - Wrap-up and open discussion
20:30 - Venue concludes (but we usually go until 10pm)
Who should apply
Developers, founders, engineers who've been building with agentic coding tools (Claude Code, Cursor, Codex, Aider, etc.) for at least a few months.
Bonus if you can share one pattern you kept and one you dropped.
Location
STARTPLATZ Cologne (exact address shared after admission)