Rung 01 Harnesses & Agent SystemsSwitch rungClose
In scope: scaffolding and orchestration, tool use, context and memory management, retries and recovery, agent evaluation and benchmarks, failure modes at step three, agent security and the exploitation surface, and cost per completed task. The load-bearing finding this rung already holds: for long-horizon agentic work the harness is often a stronger performance determinant than the model.
The Binding Constraint
Why the harness — not the model — often decides a long-horizon agent run, and what the measurements had to hold fixed to prove it
QuizIntermediate5m
Where Multi-Step Runs Actually Break
Rank agent failure modes by measured share — navigation dominates, and the best agents still miss two thirds of long-horizon tasks
Rank OrderIntermediate4m
Agentic Reasoning: Key Claims
Test your understanding of the three-layer agentic reasoning framework and its implications
QuizIntermediate5m