ControlArchitectural inferencev1.10.0
Adapter Reproduction Boundaries
Evidence levelArchitectural inference
The reproduction boundary is the control surface where adapter variants are generated, evaluated, retained, promoted, or retired. It must be explicit. If it is implicit, reproduction becomes an ordinary engineering side effect.
Minimum boundary controls
| Control | Purpose |
|---|---|
| Candidate quota | Prevents unbounded variant generation. |
| External evaluator authority | Prevents candidates from defining success. |
| Signed parent manifest | Records exactly what produced the child. |
| Composition manifest | Records the stack in which behavior was tested. |
| No-op outcome | Allows safe non-growth. |
| Synthetic-data quarantine | Prevents rejected behavior from becoming training material. |
| Memory write review | Prevents persistence through memory. |
| Promotion separation | Keeps candidate generation separate from release authority. |
| Ecological rollback packet | Restores model, adapters, route, memory, evaluator, alias, permissions, and data dependencies. |
What the boundary is not
A reproduction boundary is not a ban on all fine-tuning, adapters, or automated search. It is a separation of powers. Generation may be automated. Selection must be externally governed. Release must remain bounded by human-owned policy and reproducible evidence.
Failure mode
Evidence levelArchitectural inference
The boundary fails when a successful child can alter its evaluator, hide its lineage, write its own promotion evidence, expand permissions, or preserve rejected behavior in memory or training data.