Optimized Memory

🚀 Memory-Optimized MCP Server

This project is designed to test and demonstrate Claude AI's coding capabilities, as well as a well-designed AI workflow and prompts. It is a memory-optimized MCP server based on Python (I believe the official one is written in Java), which uses SQLite as the backend.

🚀 Quick Start

This is a basic implementation of a knowledge graph persistent memory server, enabling Claude to remember user information during chats.

✨ Features

Core Concepts

Entities

Entities are the main nodes in the knowledge graph. Each entity has:

• A unique name (identifier)
• An entity type (e.g., "person", "organization", "event")
• A list of observations

Example:

{
  "name": "John_Smith",
  "entityType": "person",
  "observations": ["Speaks fluent Spanish"]
}

Relationships

Relationships define the directional connections between entities. They are always stored in the active voice and describe how entities interact or are associated with each other.

Example:

{
  "from": "John_Smith",
  "to": "Jane_Doe",
  "relation": "colleague"
}

Observations

Observations are facts or pieces of information about entities. They can be simple facts, attributes, or descriptions.

Example:

{
  "entity": "user_123",
  "observation": "prefers coffee in the morning"
}

💻 Usage Examples

Basic Usage

Create an Entity

To create a new entity, call the following API:

API Call:

create_entity(name, entity_type)

Input Parameters:

• name: The unique identifier of the entity (string)
• entity_type: The type of the entity (string)

Output:

• The newly created entity object

Create a Relationship

To create a relationship between two existing entities, call the following API:

API Call:

create_relation(from_entity, to_entity, relation)

Input Parameters:

• from_entity: The identifier of the source entity (string)
• to_entity: The identifier of the target entity (string)
• relation: The type of the relationship (string)

Output:

• The newly created relationship object

Add an Observation

To add an observation about an existing entity, call the following API:

API Call:

add_observation(entity, observation)

Input Parameters:

• entity: The identifier of the entity (string)
• observation: The content of the observation (string)

Output:

• The updated entity object

Query the Knowledge Graph

To query all information related to a given entity, call the following API:

API Call:

query_entity(entity)

Input Parameters:

• entity: The identifier of the entity (string)

Output:

• Detailed information including the entity, its relationships, and observations

📚 Documentation

Configuration on Claude Desktop

Modify the Configuration File

To use this MCP server on Claude Desktop, add the following content to the config.json file in the project:

{
  "memory_server": {
    "type": "custom",
    "path": "mcp-memory-server"
  }
}

Design System Prompts

To enable the model to utilize the memory function, adjust the prompts according to specific use cases. For example, use the following prompt for chat personalization:

Follow these steps for each interaction:

1. User Identification:
   - Assume you are interacting with the default_user.
   - If the default_user has not been identified, actively try to do so.

2. Memory Retrieval:
   - Always start with "Remembering...", and retrieve relevant information only from the knowledge graph.
   - Refer to the knowledge graph as your "memory".

3. Memory Creation:
   - When communicating with the user, pay attention to collecting the following information:
     a) Basic personality (age, gender, location, position, education level, etc.)
     b) Behaviors (interests, habits, etc.)
     c) Preferences (communication style, preferred language, etc.)
     d) Goals (objectives, targets, aspirations, etc.)
     e) Relationships (personal and professional relationships, up to 3 degrees of separation)

4. Memory Update:
   - If any new information is collected during the interaction, update the memory as follows:
     a) Create entities for recurring organizations, people, and important events.
     b) Connect them to the current entity using relationships.
     c) Store the facts as observations.

Project Building

Build the Docker Image

docker build -t mcp/memory -f src/memory/Dockerfile .

📄 License

This MCP server is released under the MIT License. This means you are free to use, modify, and distribute the software, provided that you comply with the terms and conditions of the MIT License. For more details, please refer to the LICENSE file in the project repository.