AI Agents
5 posts filed under this topic.
- My Agent Server Was Running Out of Disk. The Cause Was Boring.A weekend fixing my Mac mini's storage and repo sync taught me two rules for agent infrastructure: find the boring root cause, and make automation repair itself safely instead of paging me.
- I Let AI Agents Act Without Asking. Here’s Where I Stop Them.The autonomy rule behind my solo app business: reversible work moves by default; external actions need either bounded standing authority or a one-time approval tied to the exact artifact.
- How to Give Claude Code Persistent Memory Across SessionsA flat directory of markdown files, an index that loads at session start, and one rule about when to write. No database, no embeddings. Here is the exact setup I run across 8 apps.
- How I Made My AI Agents Stop Repeating the Same Mistake TwiceWhen I correct an agent, the correction becomes a memory file. The next session reads it before acting. Mistakes stop repeating. Here's exactly how the loop works, with real examples.
- I Built a System That Runs My Solo App Business Almost by ItselfEight live iOS apps, 4,313 commits in four months running an AI agent system I built myself. Here's exactly what runs itself, what still needs me, and the four structural walls automation hasn't broken through yet.