Skip to main content
Every execution in Parmana follows the same deterministic lifecycle, regardless of what the underlying action is: 
Authority
    |
Authorization
    |
Intent
    |
Policy Reference
    |
Business Transaction
    |
Policy Evaluation
    |
Decision
    |
Execution Authorization
    |
Execution
    |
Execution Evidence
    |
Execution Trust Record
    |
Verification
Each stage produces an immutable artifact, and each artifact contributes to the overall trust chain — see Concepts Overview for what each one is.

​
Stage-by-stage

  1. Authority / Authorization / Intent / Policy Reference are assembled into a Business Transaction and submitted (POST /execute or client.execute()).
  2. Policy Evaluation runs the transaction’s signals against the pinned policy version, producing a Decision.
  3. If the Decision approves the intent, Execution Authorization is granted and Execution proceeds. If not, execution never runs — this is the fail-closed default.
  4. Execution produces Execution Evidence, which is folded into the Execution Trust Record.
  5. Verification can be triggered at any later point, independently, to confirm the recorded execution matches its evidence — it is not a mandatory step of the lifecycle itself, but the mechanism by which trust in a past execution is established for a third party.

​
Error handling

If validation fails at any stage before execution, no Business Transaction — and no Execution Trust Record — is created. See REST API → Error Model for the exact failure semantics and error codes.

Runtime Pipeline

The component-level implementation of this lifecycle.

Guides: Basic Execution

A concrete, working walkthrough.