Tag · 1 post

evals

Evaluating AI systems: how I define what success looks like before an agent acts, and check afterwards whether its work actually met that bar.

Command: the design decisions behind an operator that must prove its fixes

The key design decisions in Command, my Claude Code operator: a free deterministic watcher, per-repo config it writes, and a success criterion on every fix.

Jun 11, 2026
agentsclaude-code