Long-horizon, sparse-reward agent-assisted RL

A single completion is one decision. An agent-assisted coding episode is a hundred: read this file, grep that symbol, edit, run tests, read the error, edit again. The reward shows up only at the very end.

This is the charter's longer-horizon bet: extending RL to agent-assisted tasks whose episodes span many tool calls. The deep-dive may want you to reason about credit assignment when one terminal scalar has to be apportioned across dozens of them.