fix-duplicates

To merge duplicate memory entities, the skill applies deterministic merging using memory.merge_entities, prioritizing the entity with more memories as the target and recording an alias for the merged name.

You need to dedupe your memory graph when duplicates cause confusion, inconsistent histories, or broken relationships, which typically happens after data imports or during user conversations when similar entries are suspected.

Yes, you can clean up duplicate entries during conversations by explicitly requesting a dedupe or merge, which triggers the deterministic merge workflow to update records automatically.

The best way to dedupe similar names is to request a merge, which identifies the duplicates, prioritizes the entity with more memories as the target, and records the merged name as an alias.

Yes, memory management deduplication works across different entity types, automatically detecting and merging duplicates for people, projects, and organizations to maintain a clean memory graph.

After a dedupe operation, the merged entity name is recorded as an alias, ensuring the deterministic merge preserves the relationship history while keeping the memory graph clean and accurate.