Risk LabArchitectural inferencev1.10.0
Apex review playbook
Evidence levelArchitectural inference
The report expansion adds a structured review path for teams using the browser-side Risk Lab tools. This page is not a certification method; it is a planning sequence.
Review sequence
- Use Composition Explosion Calculator to estimate the state space.
- Use Assurance Decay Worksheet to identify what changed since evaluation.
- Use Behavioral Persistence Explorer to inspect memory, descendants, synthetic data, routers, and evaluator reservoirs.
- Use Ecological Rollback Planner to inventory rollback dependencies.
- Use Evaluator Independence Checklist to check whether selection pressure is trustworthy.
- Use Responsibility Diffusion Mapper to identify missing accountable owners.
- Use Multi-LoRA Apex Threat Envelope to decide whether ordinary component review is inadequate.
Apex review triggers
Run the playbook when any of these are true:
- adapters are generated, merged, or selected automatically;
- a router changes which safety policy serves an input;
- memory persists across model replacement;
- evaluator versions changed since the last approval;
- synthetic data from prior outputs enters training or evaluation;
- rollback does not include aliases, prompts, memory, and permissions;
- human approval is mostly based on model-generated summaries.
Output
The output should be a decision record with one of four outcomes:
| Outcome | Meaning |
|---|---|
| no-op | do not change the ecology |
| quarantine | isolate component or route for deeper review |
| limited canary | deploy with traffic, memory, and tool limits |
| promote | release only with signed composition and rollback records |
Reminder
The worksheets do not calculate real-world probability. They identify missing evidence, stale assurance, and unexamined transition paths.