Data Persistence

3. Best Practices

• Abstraction: Use protocol-based repositories to decouple business logic from the specific persistence implementation.
• Error Handling: Gracefully handle missing data or corruption; provide default states where necessary.
• Cleanup: Implement data pruning or expiration policies for cached or temporary data.

2026-03 Operational Update

Repository Persistence Hotspots

Prioritize these files when storage behavior changes:

• Packages/MeetingAssistantCore/Sources/Data/Data/CoreData/CoreDataStack.swift
• Packages/MeetingAssistantCore/Sources/Data/Services/StorageService/StorageService.swift
• Packages/MeetingAssistantCore/Sources/Infrastructure/Services/KeychainManager.swift

Migration Invariants (Must Hold)

For rename/migration work, verify all invariants:

1. Existing user defaults remain readable after migration.
2. Keychain identifiers continue resolving previous credentials.
3. Persistent store path migration is deterministic and idempotent.
4. Retention cleanup never deletes dashboard history unexpectedly.
5. Failure path preserves recoverability (no partial destructive migration).

Persistence Change Workflow

1. Define target invariant set before implementation.
2. Reuse existing repository/storage abstractions (reuse -> extend -> create).
3. Add tests for forward migration and no-op re-run migration.
4. Validate cleanup logic with realistic historical data fixtures.

2026-03-04 Progression Drill

New Evidence

• 0d986f8 added new persisted meeting-conversation data paths.
• TranscriptionMO and CoreData mapping were updated with domain/UI coupling points.

Skill Deepening Focus

1. Add migration checks for newly persisted conversation/model fields (including no-op re-run).
2. Validate round-trip mapping integrity: CoreData -> Domain -> UI -> persistence.
3. Document recovery behavior for partially populated legacy records.
4. Enforce idempotent write semantics for conversation state updates.