Skip to main content
Jira progress: loading…

ZAMG-DRR

Deterministic Resolution Rules

ZAMG resolves manifests using a deterministic-first model.

Resolution order

  1. explicit approved source metadata
  2. deterministic rules
  3. registry lookups
  4. template inheritance
  5. AI suggestions
  6. human review

Why this matters

This ensures that:

  • registry-backed values remain authoritative
  • AI does not silently override canonical matches
  • generation remains reproducible and auditable

Deterministic examples

  • artifact type from source path or template
  • operation key from handler name normalization
  • opcode from opcode registry
  • domain from artifact type template
  • routing mode from artifact type defaults
  • trust threshold from opcode + artifact class defaults

Review triggers

Review is required when:

  • a critical field cannot be resolved deterministically
  • a registry collision occurs
  • public exposure changes
  • ruleset resolution is ambiguous
  • CSI mapping is AI-suggested for a critical field



GitHub RepoRequest for Change (RFC)