ADR-0009 — Learner-agnostic DecisionIntent boundary¶
Status: ACCEPTED
Date: 2026-08-27
Decision¶
Rules, supervised-assisted policies and strategic RL policies converge on a common DecisionIntent contract.
Execution consumes DecisionIntent without needing to know the learner/policy family that generated it.
Consequences¶
Backtest and live execution semantics can be shared across strategy families.