defer export and fix duplication mention in CLAUDE

Author: leogdion

.github/workflows/cloudkit-sync-dev.yml

@@ -66,4 +66,4 @@ jobs:
           cloudkit-key-id: ${{ secrets.CLOUDKIT_KEY_ID }}
           cloudkit-private-key: ${{ secrets.CLOUDKIT_PRIVATE_KEY }}
           virtualbuddy-api-key: ${{ secrets.VIRTUALBUDDY_API_KEY }}
-          enable-export: 'true'
+          enable-export: 'false'  # Disabled until pagination is implemented (see issue #17)

.github/workflows/cloudkit-sync-prod.yml

@@ -40,4 +40,4 @@ jobs:
           cloudkit-key-id: ${{ secrets.CLOUDKIT_KEY_ID_PROD }}
           cloudkit-private-key: ${{ secrets.CLOUDKIT_PRIVATE_KEY_PROD }}
           virtualbuddy-api-key: ${{ secrets.VIRTUALBUDDY_API_KEY }}
-          enable-export: 'true'
+          enable-export: 'false'  # Disabled until pagination is implemented (see issue #17)

CLAUDE.md

@@ -656,13 +656,15 @@ The `main` branch requires:
 
 As noted in README.md, this is a **demonstration project** with known limitations:
 
-- No duplicate detection (repeated syncs create duplicate records)
-- No incremental sync (always fetches all data)
+- No incremental sync (always fetches all data from external sources)
 - No conflict resolution for concurrent updates
 - Limited error recovery in batch operations
+- **Export pagination**: Export only retrieves first 200 records per type (see Issue #8)
 
 These are intentional to keep the demo focused on MistKit patterns rather than production robustness.
 
+**Note on Duplicates**: The sync properly uses `.forceReplace` operations with deterministic record names (based on build numbers), so repeated syncs **update** existing records rather than creating duplicates.
+
 ## Additional Documentation
 
 For detailed guides on advanced topics, see: