Worked Examples
Each chapter of Loop Engineering includes one worked example to ground the architecture in a concrete scenario. This section hosts longer versions: full dialog transcripts, multi-step agent traces, complete diagnostic walks. Reads alongside the chapter, not as a replacement.
Examples by region
- Prefrontal — reasoning theatre cases where deliberation went thorough but the framing was wrong (Ch 4 extended).
- Working memory — full scaffolding-loop traces with NOTES.md state captured at each iteration (Ch 5 extended).
- Long-term memory — consolidation logs showing what was promoted, what stayed ephemeral, and why (Ch 6 extended).
- Dialog interface — full intent-clarification dialogs and dialog-repair transcripts (Ch 7 extended).
- Motor + sensory — agent action logs with side-effect annotations (Ch 8 extended).
- End-to-end loops — full eight-step traces with handoff annotations (Ch 10 extended).
- Failure diagnosis — extended worked diagnoses across the architecture (Ch 11 extended).
- Materiality cases — low / medium / high stakes examples (Ch 12 extended; see also materiality patterns).
- Cross-domain workflows — healthcare, legal, policy, education (Ch 13 extended).
Examples ship as Tier 2 content. Each becomes a standalone page with the in-book vignette as the entry point and the full trace as the depth.