Write tier
A per-organisation setting in ZopNight that caps which kinds of changes AI agents may make over MCP: none (the default), 1 metadata-only, 2 reversible, or 3 irreversible. It is a ceiling, not a grant: each call must also pass the caller's live role, the token's scope, and a global kill switch. Why it matters: It lets an organisation allow low-risk agent changes without also allowing destructive ones.
Read the term in context.
Each row is a lesson that uses this term. The lesson where it first appears is the canonical definition.