The ladder

Eight tutorials, in strictly increasing scale — from running one command to conducting a full multi-family experiment, and then to two external simulators. Each rung assumes the ones below it, states what you will write and what you will get, and shows the real output of everything it runs: every table on these pages was produced by executing the page, not typed.

Rung	Tutorial	You write	You get
L0	Five minutes	one command	a complete continual-learning run, read line by line
L1	A first run in code	~20 lines	the same machinery under your control — and a deliberately disappointing table
L2	Your own environment	one small file	a world that genuinely shifts, and metrics with something to measure
L3	Instrumenting to Tier 2	a component + a wrapper	composed rewards, per-component traces, a derived ceiling, a conformance pass
L4	Comparing algorithms	one sweep document	the cross-family table: families × seeds, mean ± spread
L5	A full experiment	a question and a matrix	an answer with discipline — plus a reproduced run, an exported policy, and a served one
L6	Train on a MORPHEUS world	a config	every layer at once against a live persistent world
L7	Run DiscoveryWorld	an adapter	a third-party research simulator through the same machinery

Where to enter. Never used the package: L0. Evaluating whether it fits your work: L0–L2 in an hour. You have an environment and a research question: climb to L5 — it is the page the ladder exists for. You operate a MORPHEUS deployment: L6 directly, after L0.

Everything the tutorials rely on is defined in Background and the concepts chapters; every number's meaning is in The metrics.