Agent Harness

A unified engineering harness that combines execution discipline, knowledge compounding, and product thinking. Born from 45万字 of real-world AI textbook writing + 9 production incidents.

Core Philosophy

Agent = Model + Harness. The model provides capability; the harness provides discipline.

Three layers, one workflow:

1. Challenge — Is this the right thing to build? (from Gstack)
2. Execute — Build it with engineering rigor (from Superpower)
3. Compound — Learn from what happened (from CE)

Task Complexity Auto-Grading

Before starting any task, assess complexity. This determines which workflow steps to run.