←Back
/dev/michael
michaellivs.com41 posts5 readersStale
Thoughts on building with AI agents, Claude Code, infrastructure, and the craft of software engineering. By Michael Livshits.
* Latest posts *
- [MAY.15.2026]Ergonomic coding surfaces for agents (a.k.a Code Mode)↗
- [MAY.04.2026]State of Browser Use, May 2026↗
- [MAR.26.2026]Web Search for Agents in 2026↗
- [MAR.22.2026]System reminders - how Claude Code steers itself↗
- [MAR.16.2026]Building napkin - a memory system for agents↗
- [MAR.13.2026]Reverse-engineering Claude's generative UI - then building it for the terminal↗
- [MAR.04.2026]Hello, World↗
- [MAR.04.2026]The Software Engineering Anarchist↗
- [FEB.25.2026]Knowledge distillation into skills via feedback loops↗
- [FEB.23.2026]The LLM app spectrum↗
- [FEB.22.2026]The Claw ecosystem: 12 personal agents, dissected↗
- [FEB.22.2026]Skills, forks, and self-surgery: how agent harnesses grow↗
- [FEB.10.2026]Your Eval Sucks and Nobody Is Coming to Save You↗
- [FEB.10.2026]The hard problem in multi-agent is context transfer↗
- [FEB.09.2026]Your RAG Pipeline Sucks and Nobody Is Coming to Save You↗
- [FEB.06.2026]What 16 parallel agents building a C compiler teaches about coordination↗
- [FEB.05.2026]Claude Code's Hidden Memory Directory↗
- [FEB.05.2026]Every CLI coding agent, compared↗
- [FEB.02.2026]A thousand ways to sandbox an agent↗
- [JAN.30.2026]The architecture behind Claude Code's $1B run-rate↗
Page 1 / 3 · 41 total