mdm-engine

Source: GitHub repo. Proposal generation runtime; produces Proposal from state/context.

Role

Proposal generation core. Data flow: (state, context) → propose() → Proposal → PacketV2.

Components

Proposal Generation

propose(features, context) → Proposal. Generic feature dict, confidence, decision-schema conforming.

Feature Extraction

build_features(event, history, ...) → dict. Domain-agnostic.

Event Loop

run_loop(source, executor, ...) — event → features → proposal → modulation → execution → trace; emits PacketV2.

Private Hook

mdm_engine/mdm/_private/model.py: compute_proposal_private(features, **kwargs) → Proposal.

Import

from mdm_engine.mdm.decision_engine import DecisionEngine