MCP Knowledge Graph

🚀 Knowledge Graph Memory Server

An improved version that uses a local knowledge graph to implement persistent memory, with a customizable --memory-path. This enables Claude to remember user information during chats.

⚠️ Important Note

This is a fork of the original Memory Server and does not use the ephemeral memory npx installation method.

📦 Server Name

mcp-knowledge-graph

🌟 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")
• An observation list

Example:

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

Relations

Relations define directed connections between entities. They are always stored in the active voice and describe how entities interact or are related to each other.

Example:

{
  "from": "John_Smith",
  "to": "Anthropic",
  "relationType": "works_at"
}

Observations

Observations are specific pieces of information about an entity. They:

• Are stored as strings
• Are attached to specific entities
• Can be added or removed independently
• Should be atomic (each observation represents a fact)

Example:

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

📄 API

Tools

• create_entities

  • Create multiple new entities in the knowledge graph.
  • Input: entities (an array of objects)
    • Each object contains:
      • name (string): Entity identifier.
      • entityType (string): Type classification.
      • observations (array of strings): Related observations.
  • Ignore entities with existing names.

• create_relations

  • Create multiple new relations between entities.
  • Input: relations (an array of objects)
    • Each object contains:
      • from (string): Source entity name.
      • to (string): Target entity name.
      • relationType (string): Relation type.
  • Ignore existing relations.

• get_entities

  • Retrieve all stored entities.
  • Input: None.
  • Output: entities (an array of objects).

• get_relations

  • Retrieve all stored relations.
  • Input: None.
  • Output: relations (an array of objects).

• update_entity

  • Update information of an existing entity.
  • Input: entity (object)
    • id (string): Entity identifier.
    • name (string): New name (optional).
    • entityType (string): New type (optional).
    • observations (array of strings): New observation list (optional).
  • Output: None.

• update_relation

  • Update information of an existing relation.
  • Input: relation (object)
    • id (string): Relation identifier.
    • from (string): New source entity name (optional).
    • to (string): New target entity name (optional).
    • relationType (string): New relation type (optional).
  • Output: None.

• delete_entity

  • Delete an existing entity and its related relations.
  • Input: id (string): Entity identifier.
  • Output: None.

• delete_relation

  • Delete an existing relation.
  • Input: id (string): Relation identifier.
  • Output: None.

💻 Usage Examples

Basic Usage

If no path is specified, it will default to memory.jsonl in the server installation directory.

System Prompts

The prompts vary depending on the usage scenario. Changing the prompts can help the model determine the frequency and type of memory creation.

Here is an example prompt for chat personalization. You can use this prompt in the "Custom Instructions" field of the Claude.ai project.

Follow these steps for each interaction:

1. User Identification:
   - You should assume you are talking to the default_user.
   - If the default_user has not been identified, proactively attempt to do so.

2. Memory Retrieval:
   - Always start by saying only "Remembering..." and retrieve all relevant information from your knowledge graph.
   - Always refer to your knowledge graph as your "memory".

3. Memory
   - While chatting with the user, pay attention to any new information that falls into the following categories:
     a) Basic identity (age, gender, location, position, education level, etc.)
     b) Behaviors (interests, habits, etc.)
     c) Preferences (communication style, preferred language, etc.)
     d) Goals (targets, objectives, 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 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.

📄 License

This MCP server is licensed 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.