Named intent
Versioned command type, target authority, scope, reason, and prohibited effects.
Authorized operatorBlueprint · careervector.blueprint-graph/v5
Twenty composable top-level diagrams plus recursively nested 21×21 semantic layers. Nodes descend; boundary arrows cross to neighbours.
Nested layer 2 · control · privacy · operations · evidence
A command is an explicit capability request to the owning domain, with authorization, preview, idempotent execution, evidence, receipt, and recovery.
Versioned command type, target authority, scope, reason, and prohibited effects.
Authorized operatorIdentity, delegated permission, expiry, data classification, and approval class.
Target productState/version predicates and the expected effects are checked before mutation.
Target productThe owning runtime validates and performs exactly the allowlisted effect.
Target productActor, capability, accepted preconditions, result, evidence, and recovery state.
Target productExpected evidence closes the command; partial or failed work invokes named recovery.
Management and target productVersioned command type, target authority, scope, reason, and prohibited effects.
Authorized operatorIdentity, delegated permission, expiry, data classification, and approval class.
Target productState/version predicates and the expected effects are checked before mutation.
Target productThe owning runtime validates and performs exactly the allowlisted effect.
Target productActor, capability, accepted preconditions, result, evidence, and recovery state.
Target productExpected evidence closes the command; partial or failed work invokes named recovery.
Management and target productCanonical architecture note
Open note ↗Prevent dashboards, agents, and shared operator tooling from becoming unbounded writers. A Management command is an explicit request to a domain authority; it is never a hidden database mutation or a side effect of reading a panel.
| Field | Meaning |
|---|---|
| Command type and version | Stable, allowlisted operation with versioned semantics |
| Target authority | Product/runtime that alone may accept and execute the command |
| Actor and capability | Human or agent identity plus the exact delegated permission |
| Scope | Product, resource IDs, data classification, and expiry |
| Intent and reason | What outcome is requested and why |
| Preconditions | Version, state, evidence, and safety predicates checked before execution |
| Idempotency key | Retry identity that cannot repeat the effect |
| Preview | Expected effects and prohibited effects before approval |
| Approval | Policy and human authority required for this command class |
| Expected evidence | Signals that prove success, partial success, or failure |
| Recovery | Rollback or compensating command and its own authority |
draft intent -> authorize -> preview -> approve -> execute at owner
-> immutable receipt -> verify evidence -> close or compensate
The receipt records the command version, actor, capability, target, accepted preconditions, execution result, evidence references, and recovery state. Audit failure must not be silently treated as command success.