Skip to main content

​
MemMachine Configuration Wizard

The MemMachine Configuration Wizard is an interactive tool designed to simplify the setup and configuration process of MemMachine. It guides users through a series of prompts to gather necessary information and automatically generates a configuration file (cfg.yml) based on the provided inputs.

​
Getting Started

First of all, install the MemMachine package if you haven’t already: 
pip install memmachine-server
Next, run the configuration wizard using the following command: 
memmachine-configure

​
Configuration Steps

The wizard will prompt you for various configuration options, including:
  • Install Neo4j: Option to install and configure a Neo4j database for memory storage.
  • Or Use Existing Neo4j: Provide connection details for an existing Neo4j instance.
  • Large Language Model (LLM) Provider: Choose from supported providers like OpenAI, AWS Bedrock, or Ollama.
  • Model Selection: Select the specific LLM and embedding models you wish to use.
  • API Keys and Credentials: Input necessary API keys or credentials for the selected LLM provider.
  • LLM Base URL: Specify the base URL for the LLM Model API if applicable.
  • Embedding Model Configuration: Configure settings related to the embedding model.
  • AWS Bedrock Settings: If AWS Bedrock is selected, provide AWS credentials and region information.
  • Server Settings: Configure server-related settings such as host, port
The wizard only supports Neo4j as the vector database at the moment. The relational database is configured to use SQLite by default. For users who wish to use other vector or relational databases, manual configuration of the cfg.yml file will be necessary after running the wizard. By default, the configuration file will be available in <HOME>/.config/memmachine/cfg.yml. Note:
  • The wizard will install Neo4j together with Java natively when the user chooses y for “install Neo4j”.
  • The wizard will NOT install large language model such as Ollama. User could either install them manually or use cloud-based LLM services such as OpenAI or AWS Bedrock.
Platform-specific behavior:
  • Windows: Downloads and extracts the ZIP distributions of both Java and Neo4j.
  • macOS: Installs Java and Neo4j via brew.
  • Linux: Downloads and extracts the tar.gz distributions of Java and Neo4j.

​
Completing the Setup

Once all prompts are answered, the wizard will generate a cfg.yml file in <HOME>/.config/memmachine/cfg.yml by default. This file contains all the configuration settings needed to run MemMachine. You can now start the MemMachine server with the generated configuration: 
memmachine-server