MCP server · Notes
Graphiti Local MCP server
Give your AI a local memory of project facts it can look up, with you approving anything new it learns.
Graphiti Local is a memory helper for your AI that runs on your own computer. It stores facts about your projects in a small local database, and your AI can look things up from it when you ask. It is handy if you want your AI to remember decisions and details without sending your notes to a cloud service.
What is an MCP server? The 30-second version
On its own, your AI can only chat with you. An MCP server is a small helper program that gives your AI a new skill or a connection to something else. This one connects your AI to a local memory store called Graphiti Local, so it can search your saved project facts and answer questions about them. You stay in control: your AI can read from the memory, but anything new it wants to add waits for your approval.
What this MCP server does
You ask your AI a question about your project, like which database a team uses. Your AI passes that question to this helper. The helper searches a local knowledge graph stored on your computer and finds the matching facts. It sends those facts back, and your AI turns them into a normal answer in the chat. If your AI suggests a new fact to remember, it goes into a separate queue, and nothing changes in the memory until you review and approve it.
What you can do with it
- Look up project decisions and facts your AI has stored
- Ask historical questions to see how a fact changed over time
- Feed in a Markdown file, text file, or folder of notes to build memory
- Review proposed facts and mark them accept, refuse, or contested
- Run a quick demo with a small sample graph and no model downloads
- Keep separate memory files for different people or teams
- Check the health of your local setup with a doctor command
Try asking your AI
- “Which database does Aurora Analytics use?”
- “What did we decide about the login flow last month?”
- “Show me the history of our choice for the main database”
- “Add these meeting notes to my project memory”
What it gives back to you
You get a short answer in the chat, based on facts pulled from your local memory. The helper can also show you the facts it found, along with when each fact was valid, so you can check them. When you review proposed updates, you see a list of facts to accept, refuse, or mark as contested. If you ingest a file, you get a receipt showing what landed and what still needs attention.
Before you start
What you need
- uv installed on your computer (a tool for running Python programs)
- Ollama installed and running, with the qwen2.5:7b and nomic-embed-text models pulled
- Internet access during the first setup to download things
- A terminal where you can run a few commands
Good to know
Your AI can only read from this memory, but the tool can change or delete things when you run commands yourself, and the model that extracts facts from your text can sometimes get things wrong, so check the facts and their dates before trusting them.
Install it with your AI
Add Graphiti Local MCP server to your AI, no technical skills needed
You don't install anything by hand. You copy one prompt, paste it into an AI that can work on your computer, and it checks, installs and connects the server for you, asking you when it needs something.
Sign in to get the install prompt
Members get a ready-made prompt that lets the Claude desktop app check Graphiti Local MCP server, install it and connect it for them, step by step. You don't need any technical skills: you copy, paste and answer a few questions. Your connected AI can also find and install any of the 4,609 MCP servers here for you.
Who it's for
People who want their AI to remember project facts locally and prefer to approve anything new before it is saved, such as developers, researchers, and small teams.
