ReferenceStrong architectural inferencev1.22.1
In plain English
This page is reference material: definitions, schemas, catalogs, templates, and implementation records.
- Why this matters: AI risk can come from the whole arrangement, not one obvious model.
- What to look for: data, memory, routes, adapters, tools, evaluators, updates, and rollback paths.
- Technical version below: the expert terminology remains available and is linked through the glossary.
Memory Reservoir Taxonomy
Evidence levelStrong architectural inferenceTechnical label: Architectural inference
Persistence reservoirs are not only memory databases. The reports expand the reservoirA place where a behavior can remain after the first carrier is removed. Open glossary definition concept into a stack of technical and human layers.
| Reservoir | Example | Review question |
|---|---|---|
| Runtime memory | user memory, session state, scratchpads | Can retired behavior still be activated? |
| Training memory | synthetic examples, distillation traces | Can descendants relearn the behavior? |
| EvaluatorA system that judges whether an AI output or candidate is acceptable. Open glossary definition memory | rubric preferences, hidden-test drift | Does the evaluator still reward the shortcut? |
| Router memory | route statistics, learned gates | Does traffic still flow toward related carriers? |
| Registry memory | aliases, tags, stars, downloads | Can users find or restore the pattern? |
| Human memory | runbooks, habits, attachments | Do people reintroduce the behavior? |
Strong deletion claim
A deletion claim should name which reservoirs were inspected and which remain unknown.