Glossary
Plain definitions of the terms used in Cogna8.
| Accountable executive | The named person answerable for an AI system. |
| Action gate | Where Cogna8 decides whether an agent may take a proposed action. |
| AI system | Any model, agent, endpoint, copilot or MCP server you run or rely on. |
| Canonical key | The name of the thing a fact is about, for example claim.amount. |
| Conflict | Two sources disagreeing about the same key in the same scope. |
| Control | A specific measure that satisfies an obligation. |
| Copilot | The screen-bound assistant that proposes changes for your approval. |
| Decision receipt | The record of one gate decision: inputs, rule and outcome. |
| Deterministic decision | A decision that does not depend on a model judgement and can be replayed. |
| Enforcement boundary | The line between model-assisted steps and rule-based decisions. |
| Evidence | A dated, sourced record that a control operated. |
| Governed state | Facts an agent relies on, held as typed, versioned items with their source. |
| Integrity score | An advisory 0 to 100 signal of how ready an action is. |
| MCP | Model Context Protocol, a standard way for agents to reach tools and data. |
| Obligation | A requirement from a framework or law, with its source and date. |
| Policy | A declarative rule for what an action needs before it can run. |
| Proof level | How far a control is proven: declared to effective. |
| Scope | The boundary that decides which state a conversation or agent can see. |
| Snapshot | A point-in-time capture of governed state. |
| State item | One governed fact, with key, value, type, status, source and version. |
Last reviewed October 2026Suggest a change