Section B• Last verified November 2026
Understanding Objectives
Translating natural language intent into structured DAG plans linked to authorized mandates.
What is an Objective?
An Objective represents a high-level business or research goal submitted by a user. While a mandate defines the financial ceiling, the objective defines what work should be performed, how it is broken down into tasks, and when it is considered successfully completed.
Natural Language to Structured Plan
Users enter goals in plain English. Mandate's planning orchestrator converts the text into a Directed Acyclic Graph (DAG) with explicit task dependencies, required agent capabilities, and a deterministic SHA-256 plan hash.
Generated DAG Plan Structure
Independent Success Criteria
Objectives do not complete based on self-reported agent statements. Completion requires independent verification:
- All dependent DAG tasks must have reached verified terminal states.
- All financial disbursements must have confirmed onchain transaction receipts on NEAR.
- Verification evidence must match expected output fingerprints.
Objective Lifecycle
Diagram 3: Multi-Agent DAG Objective Lifecycle
Anti-Self-Certification1
Natural Language GoalUser defines objective & links active mandate2
Structured DAG PlanTasks, dependencies & deterministic plan hash3
User Plan ApprovalHuman reviews DAG before any execution begins4
Capability MatchingAgents leased via capability registry5
Task ExecutionTools & integrations run within policy bounds6
Independent VerificationOutputs verified via system fingerprint proofs7
Objective CompletionCompleted only when all criteria independently pass