https://ottto.net/blog/ Notes from building agent-native engineering at Ottto. en https://ottto.net/blog/agent-memory-audit/ https://ottto.net/blog/agent-memory-audit/ Claude Code quietly writes notes to itself about your project - what broke, what you prefer, what it learned the hard way. We audited every note our agents wrote across two machines, measured what memory actually costs, and found the real problem isn't the one everyone talks about. The expensive part isn't the diary. It's that the diary and the documentation stopped agreeing. Fri, 17 Jul 2026 00:00:00 GMT https://ottto.net/blog/landing-lander/ https://ottto.net/blog/landing-lander/ Our agents were opening pull requests faster than we could merge them, and the queue just kept getting bigger. So we made landing itself an agent - a skill running on a /loop that batches pull requests, lands them, and supervises itself. It put 32 PRs through full CI and deploy in two hours, without skipping a single check. The skill is open source, and you can run it yourself. Mon, 06 Jul 2026 00:00:00 GMT