occupation
unclassified
description
Evaluate the coding harness you are running inside (pi, opencode, claude-code) on its live setup and current model, via bench setup run. Measures it with and without its skills, MCP servers and plugins on --ab, and reports which of them the run actually…
updated