The MCP Knowledge Graph server provides persistent memory for AI models by implementing a local knowledge graph that allows models to remember information across conversations. This server works with any AI model supporting the Model Context Protocol (MCP) or function calling capabilities.
To use the MCP Knowledge Graph server, you need to configure your AI platform to connect to it. The server is available via NPX.
Add this configuration to your claude_desktop_config.json file:
{
"mcpServers": {
"memory": {
"command": "npx",
"args": [
"-y",
"mcp-knowledge-graph",
"--memory-path",
"/path/to/your/memory.jsonl"
],
"autoapprove": [
"create_entities",
"create_relations",
"add_observations",
"delete_entities",
"delete_observations",
"delete_relations",
"read_graph",
"search_nodes",
"open_nodes"
]
}
}
}
Make sure to replace /path/to/your/memory.jsonl with your preferred storage location. If no path is specified, the server will default to memory.jsonl in its installation directory.
Entities are the primary nodes in the knowledge graph. Each entity has:
Example entity:
{
"name": "John_Smith",
"entityType": "person",
"observations": ["Speaks fluent Spanish"]
}
Relations define directed connections between entities in active voice.
Example relation:
{
"from": "John_Smith",
"to": "ExampleCorp",
"relationType": "works_at"
}
Observations are discrete pieces of information about an entity:
Example:
{
"entityName": "John_Smith",
"observations": [
"Speaks fluent Spanish",
"Graduated in 2019",
"Prefers morning meetings"
]
}
Creates multiple new entities in the knowledge graph.
{
"entities": [
{
"name": "John_Smith",
"entityType": "person",
"observations": ["Speaks Spanish", "Lives in New York"]
},
{
"name": "ExampleCorp",
"entityType": "organization",
"observations": ["Founded in 2010"]
}
]
}
Creates new relations between entities.
{
"relations": [
{
"from": "John_Smith",
"to": "ExampleCorp",
"relationType": "works_at"
}
]
}
Adds new observations to existing entities.
{
"observations": [
{
"entityName": "John_Smith",
"contents": ["Enjoys hiking", "Has a dog named Max"]
}
]
}
Removes entities and their relations.
{
"entityNames": ["John_Smith"]
}
Removes specific observations from entities.
{
"deletions": [
{
"entityName": "John_Smith",
"observations": ["Enjoys hiking"]
}
]
}
Removes specific relations from the graph.
{
"relations": [
{
"from": "John_Smith",
"to": "ExampleCorp",
"relationType": "works_at"
}
]
}
Reads the entire knowledge graph. No input required.
Searches for nodes based on a query.
{
"query": "Smith"
}
Retrieves specific nodes by name.
{
"names": ["John_Smith", "ExampleCorp"]
}
Here's an example prompt for chat personalization that you can adapt for any AI model:
Follow these steps for each interaction:
1. User Identification:
- You should assume that you are interacting with default_user
- If you have not identified default_user, proactively try to do so.
2. Memory Retrieval:
- Always begin your chat by saying only "Remembering..." and retrieve all relevant information from your knowledge graph
- Always refer to your knowledge graph as your "memory"
3. Memory Gathering:
- While conversing with the user, be attentive to any new information that falls into these categories:
a) Basic Identity (age, gender, location, job title, education level, etc.)
b) Behaviors (interests, habits, etc.)
c) Preferences (communication style, preferred language, etc.)
d) Goals (goals, targets, aspirations, etc.)
e) Relationships (personal and professional relationships up to 3 degrees of separation)
4. Memory Update:
- If any new information was gathered during the interaction, update your memory as follows:
a) Create entities for recurring organizations, people, and significant events
b) Connect them to the current entities using relations
c) Store facts about them as observations
For Claude users, you could use this prompt in the "Custom Instructions" field of a Claude.ai Project. For other models, adapt it to their respective instruction formats.
To integrate with any AI model that supports function calling:
The server's knowledge graph structure and API are model-agnostic, allowing for flexible integration with various AI platforms.
There are two ways to add an MCP server to Cursor. The most common way is to add the server globally in the ~/.cursor/mcp.json file so that it is available in all of your projects.
If you only need the server in a single project, you can add it to the project instead by creating or adding it to the .cursor/mcp.json file.
To add a global MCP server go to Cursor Settings > MCP and click "Add new global MCP server".
When you click that button the ~/.cursor/mcp.json file will be opened and you can add your server like this:
{
"mcpServers": {
"cursor-rules-mcp": {
"command": "npx",
"args": [
"-y",
"cursor-rules-mcp"
]
}
}
}
To add an MCP server to a project you can create a new .cursor/mcp.json file or add it to the existing one. This will look exactly the same as the global MCP server example above.
Once the server is installed, you might need to head back to Settings > MCP and click the refresh button.
The Cursor agent will then be able to see the available tools the added MCP server has available and will call them when it needs to.
You can also explictly ask the agent to use the tool by mentioning the tool name and describing what the function does.