Packages

Each package documents its own purpose, dependency rules, public API, extension guide, design rationale, and anti-patterns. These pages reuse the packages' own README.md files verbatim - the same documentation enforced by each package's test_public_api.py.

For the symbol-level, docstring-generated reference, see the API Reference.

Foundation

Package	Responsibility
core	Framework primitives: entities, value objects, identifiers, specifications, repositories, builders.
ontology	The typed, machine-readable domain vocabulary (ten sub-ontology families) and the Knowledge Graph projection.
events	The immutable, append-only event model (Event Sourcing): store, bus, replay/time-travel, codecs.
registry	The generic, type-safe registration/discovery mechanism every domain registry specializes.
plugins	The plugin lifecycle layer built on registry: manifests, states, activation ordering.
connectors	The vendor-neutral ingestion boundary: the FMSConnector contract and reference connectors.
kpis	The metadata-first, self-describing KPI Engine and the Standard Library.

Intelligence

Package	Responsibility
analytics	Statistical and analytical computation built on kpis.
decision	The prescriptive layer: rules, policies, ranking, explanation, alerting, audit.
digital_twin	The stateful representation layer: Twin, synchronization, snapshots, telemetry.
simulation	The projection layer: scenarios, runs, experiments, methodology extension points.
optimization	The prescriptive-search layer: six solving paradigms, plan comparison, sensitivity.
agents	The model-independent orchestration layer: tasks, policies, approval, delegation, audit.
visualization	The presentation layer: dashboards, reports, the rendering pipeline - the final package.