In my agent fleet, role moves the metrics more than the model does — and the role mix differs per model, so pooled comparisons measure the dispatcher.
Sub-Agent Metrics Are Not Comparable to Main-Thread Metrics
Related
I built a Skill and checker for MCP's breaking change 2026-08-26. Then the checker was wrong about it.
A package rename looks exactly like a package that doesn't exist. My tool got it wrong, my fix got it wrong in the other direction, and the lesson generalises further than MCP. Plu...
What an AI Engineer job actually tests
An AI Engineer job posting almost never asks you to train a model. It asks for Python. Backend services. APIs. Deployment. Say you are two years into…
Designing Edit Operations for AI Agents
Four lessons from building IWE's block-editing language for LLM writers: state the blast radius, make...