openmemory: failed memory deletion with qdrant point vector_id as UUID

[stev-tig]

🐛 Describe the bug

Hit the problem when I'm using openmemory-mcp, trying to trigger delete memories
in the configuration, it's using qdrant as vector store.

error log as follows

openmemory-mcp    |                     INFO     Deleting memory with                    main.py:902
openmemory-mcp    |                              memory_id=UUID('15f0fe7f-8581-4ded-bdaf            
openmemory-mcp    |                              -60e0808b7b04')                                    
openmemory-mcp    |                     WARNING  Failed to delete memory           mcp_server.py:340
openmemory-mcp    |                              15f0fe7f-8581-4ded-bdaf-60e0808b7                  
openmemory-mcp    |                              b04 from vector store: 2                           
openmemory-mcp    |                              validation errors for                              
openmemory-mcp    |                              PointRequest                                       
openmemory-mcp    |                              ids.0.int                                          
openmemory-mcp    |                                Input should be a valid integer                  
openmemory-mcp    |                              [type=int_type,                                    
openmemory-mcp    |                              input_value=UUID('15f0fe7f-8581-4                  
openmemory-mcp    |                              ded-bdaf-60e0808b7b04'),                           
openmemory-mcp    |                              input_type=UUID]                                   
openmemory-mcp    |                                  For further information visit                  
openmemory-mcp    |                              https://errors.pydantic.dev/2.11/                  
openmemory-mcp    |                              v/int_type                                         
openmemory-mcp    |                              ids.0.str                                          
openmemory-mcp    |                                Input should be a valid string                   
openmemory-mcp    |                              [type=string_type,                                 
openmemory-mcp    |                              input_value=UUID('15f0fe7f-8581-4                  
openmemory-mcp    |                              ded-bdaf-60e0808b7b04'),                           
openmemory-mcp    |                              input_type=UUID]                                   
openmemory-mcp    |                                  For further information visit                  
openmemory-mcp    |                              https://errors.pydantic.dev/2.11/                  
openmemory-mcp    |                              v/string_type                                      

qdrant accepts id as int64 or UUID string, however when triggering _delete_memory in main.py, from the logs above it mentioned the input_value failed validation

from the stack trace main.py:902 it's probably in this section of code

    def _delete_memory(self, memory_id):
        logging.info(f"Deleting memory with {memory_id=}")
        existing_memory = self.vector_store.get(vector_id=memory_id)
        prev_value = existing_memory.payload["data"]
        self.vector_store.delete(vector_id=memory_id)
        self.db.add_history(
            memory_id,
            prev_value,
            None,
            "DELETE",
            actor_id=existing_memory.payload.get("actor_id"),
            role=existing_memory.payload.get("role"),
            is_deleted=1,
        )
        capture_event("mem0._delete_memory", self, {"memory_id": memory_id, "sync_type": "sync"})
        return memory_id

the line self.vector_store.delete(vector_id=memory_id) probably is triggering qdrant.py:154

    def delete(self, vector_id: int):
        """
        Delete a vector by ID.

        Args:
            vector_id (int): ID of the vector to delete.
        """
        self.client.delete(
            collection_name=self.collection_name,
            points_selector=PointIdsList(
                points=[vector_id],
            ),
        )

[mimi3421]

Similar problem also occurs when _add_memories is called. Although this error may not affect the actual database modification, the log in the MCPhost should be somting like this:

Calling openmemory__add_memories
┃ Arguments: "text":"I have a dog in the cage, it's called John."
┃  Tool Call (28 Jun 2025 05:18 PM)

┃ openmemory__add_memories: {"text":"I have a dog in the cage, it's called John."}
┃ {"content":[{"type":"text","text":"Error executing tool add_memories: 1 validation error for add_memoriesOutput\nresult Input should be a valid string [type=string_type, input_value={'results': [{'id': 'b070...John\", 'event': 'ADD'}]}, input_type=dict] For further information visit https://errors.pydantic.dev/2.11/v/string_type"}],"isError":true}

I can't locate where the pydantic validation is called, as the pure call of memory_client.add like what is done in openmemory/api/app/mcp_server.py below triggers NO error.

            user, app = get_user_and_app(db, user_id=uid, app_id=client_name) // line 80
            response = memory_client.add(text,
                                         user_id=uid,
                                         metadata={
                                            "source_app": "openmemory",
                                            "mcp_client": client_name,
                                        }) # line 86

Maybe the error occurs at the modification of the local sqlite database openmemory.db after the interaction with qdrant.