Forge Intelligence

Migration from generic knowledge modules to native MathGenesis objects

The current generic module shape is suitable for formulas, decision tables, examples, exceptions, and time-varying facts. MathGenesis needs stronger distinctions among axioms, definitions, theorems, proof receipts…

Updated

Why migrate

The current generic module shape is suitable for formulas, decision tables, examples, exceptions, and time-varying facts. MathGenesis needs stronger distinctions among axioms, definitions, theorems, proof receipts, algorithms, constructions, physical models, partitions, and epistemic states.

Phases

Phase 0 — Compatibility wrapper

  • Keep the existing schema and registry.
  • Put native objects inside a mathgenesis_native_bundle record in formulas.
  • Validate both the wrapper and the nested native objects.

Phase 1 — Native registry

  • Add knowledge/math/registry.yaml and schemas/mathgenesis/.
  • Add object, source, symbol, dependency, and snapshot indexes.
  • Continue emitting compatibility wrappers for old consumers.

Phase 2 — Native retrieval

  • Route math, geometry, and physics packs directly to the native registry.
  • Enforce partition-scoped retrieval credentials.
  • Add dependency-aware and notation-aware retrieval.

Phase 3 — Discovery lifecycle

  • Add staging, falsifier, validation, utility, contamination, promotion, and historian artifacts.
  • Store generation snapshots and immutable lineage.

Phase 4 — Deprecate compatibility reads

  • Keep wrapper exports for archival compatibility.
  • Make native objects canonical.
  • Publish a migration ledger and expiry policy for legacy consumers.