What problem does it solve? Teams running multiple release trains need a controlled, auditable way to move fixes and features between trains without manual git cherry-pick, and the bubbles.propagate operator refuses to run until a valid propagation-policy.yaml contract exists. ## Core Features & Use Cases - Policy Schema Authoring: Defines the v1 schema for propagation-policy.yaml, including trains, defaultFlow edges, validation modes, and backport settings. - Validation Depth Control: Explains validate-only, full-delivery, and none modes so each receiving train gets the appropriate gate enforcement. - Backport Governance: Documents the approval-token flow and backportable edge flags required before any backport operation is permitted. - Use Case: A team with experimental, mvp, and prod trains writes a policy declaring experimental-to-mvp as automatic with validate-only, and mvp-to-prod as manual with full-delivery, then audits drift safely before any mutation. ## Quick Start Copy templates/propagation-policy.yaml.tmpl to propagation-policy.yaml, edit the trains and defaultFlow edges to match config/release-trains.yaml, then run the propagation-policy-guard.sh script to validate it.