Choose a complete operational walkthrough

Each route starts from a saved runnable script with its setup included. Read the named prerequisite guides first, or open the complete example directly.

Outcome	Start with	What to inspect
Build a working application from generated classes	Codegen, Expert 36	Word counts, object state, unique reuse, independent workers
Compare permitted and rejected code	Agent rooms, Expert 07	Acceptance payloads and named validation issues
Prepare a candidate without changing the live version	Governed change, Expert 22	Unchanged heads and original pricing behavior
Reconcile source across two modules	Compositions, Expert 30	Replaced/added parts and returned candidate source
Preserve a record across runtime teardown	Restore, Expert 27	Local cache survival and the actual load report or refusal
Round-trip a research record as JSON	Research, Expert 28	Preserved set identity and lane organization
Run several rooms concurrently	Concurrency, Expert 21	Exact job results and distinct workstations

A script pass establishes its executed assertions and handled outcomes. For an operational deployment, also inspect the application-specific result: a reported refusal, a local checkpoint, and a completed remote restore are different outcomes.

Runnable examples

An agent builds a working system — Expert 36
The codegen room agents writing code — Expert 07
Asking what code would do before running it — Expert 22
Composing a version in the workshop — Expert 30
A world that outlives its own runtime — Expert 27
The record crosses as a json string — Expert 28
Many agents writing code at once — Expert 21

All expert examples · Level contents · Full contents

Canonical page source