Writing
Essays and field notes
Practical notes on the systems around models: protocols, harnesses, runtimes, evals, and the engineering choices that make AI useful.
It's Not the LLM. It's the Harness.
Why the same model behaves like a different agent when context, tools, state, permissions, and control loops change around it.
Agents / Harnesses / Infrastructure
It's Not the LLM. It's the MCP.
Why capable agents fail when their tool interfaces overwhelm context, and how small verbs plus registry dispatch change the architecture.
MCP / Agents / Infrastructure