@rapidclaw

[Snapshot, Diff, Rollback] What Agent State Looks Like After 14 Days In a MicroVM

2h ago · 6 min read · When you let an AI agent run for two weeks straight, "state" stops meaning what you think it means. On day one, the agent's state is roughly what you'd expect: a prompt context, a tool-call history, m

5 ways your AI agent runtime silently dies overnight (and the boring fix for each)

May 4 · 7 min read · I ran the same agent for thirty straight days. It died five times. Four of them did not show up in any log I had set up ahead of time, which is the part that bothers me most. By the end I had a check

MicroVM vs Docker for AI agents: I gave one sudo and broke the other

May 4 · 8 min read · Last week I ran a small experiment that I should have run a year ago. Same agent code. Same model. Same task list: install three Python packages from a CSV, fetch a few APIs, write JSON to disk, run

[Observability] for Long-Running AI Agents: A Field Guide to What to Log

Apr 30 · 5 min read · The first time we ran an AI agent for 30 days straight, we logged nothing useful. The agent worked. Then the agent stopped working. We had no idea why. That's the typical observability story for AI ag

[The Silent Tax] Why Your AI Agent's Tool Calls Need a Budget, Not a Prayer

Apr 23 · 6 min read · If you've ever watched an AI agent loop through twenty tool calls to answer a question that should have taken three, you already know what this article is about. The dirty secret of agentic systems in