Optimized Memory MCP Server

🚀 Optimized Memory MCP Server

This project is designed to test and demonstrate the programming capabilities of Claude AI, along with a well - designed AI workflow and prompt design. It is a Python - based fork of the local knowledge graph memory MCP server (the official implementation is believed to be in Java), using SQLite as the backend.

🚀 Quick Start

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

✨ Features

Core Concepts

Entities

Entities are the primary 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"]
}

Relations

Relations 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": "Anthropic",
  "relationType": "works_at"
}

Observations

Observations are discrete pieces of information about entities. They are:

• Stored as strings
• Attached to specific entities
• Can be added or removed independently
• Should be atomic (each observation represents a single fact)

Example:

{
  "entityName": "John_Smith",
  "observations": [
    "Speaks fluent Spanish",
    "Graduated in 2019",
    "Likes morning meetings"
  ]
}

💻 Usage Examples

Basic Usage

In the beginning of a chat, Claude will actively retrieve all relevant memories and prompt the user with "Recalling...". Throughout the conversation, Claude will collect new information and update it in the knowledge graph.

Advanced Usage

Prompt Design

It is recommended to adjust the prompt content according to specific use cases so that the model can decide the frequency and type of memory. Here is an example prompt for chat personalization:

Follow these steps for each interaction:

1. User Identification:
   - Assume you are talking to the default_user.
   - If the default_user has not been identified, actively try to identify them.

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

3. Memory Content:
   - When communicating with the user, pay attention to collecting new information in the following categories:
     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 three degrees of separation)

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

📦 Installation

Use Docker to build the server:

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

📚 Documentation

API

Tools

Entity Management

• getEntities(): Retrieve all entities.
• addEntity(entity): Add a new entity.
• updateEntity(entityId, updates): Update the attributes of a specified entity.

Relation Management

• getRelations(): Retrieve all relations.
• addRelation(relation): Add a new relation.
• removeRelation(relationId): Remove a specified relation.

Observation Management

• getObservations entityId: Retrieve all observations of a specified entity.
• addObservation(entityId, observation): Add a new observation to a specified entity.
• updateObservation(observationId, updates): Update a specified observation.

📄 License

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