postg-mem
5
PostgMem is a .NET service designed to store and manage AI memories using vector embeddings in PostgreSQL. It supports semantic search and tagging functionalities, and integrates seamlessly with the Model Context Protocol for AI applications.
PostgMem
PostgMem is a .NET-based service for AI agents offering memory storage and retrieval through vector embeddings, leveraging PostgreSQL and the pgvector extension for effective similarity search.
Key Features
- Store and retrieve structured memories with vector embeddings
- Semantic search through memories using vector similarity
- Filter results using tags
- Integrates with MCP for AI agent ease
Technologies
- .NET 9.0
- PostgreSQL with pgvector
- Model Context Protocol
- ASP.NET Core
Prerequisites
- .NET 9.0 SDK
- PostgreSQL with pgvector
Running the Application
Run the server with dotnet run to start the MCP server at http://localhost:5000 by default.
MCP Tools
- Store: Add a new memory
- Search: Find similar memories
- Get: Retrieve by ID
- Delete: Remove by ID