Eidetic Semantic Memory Server

Example Use Cases

• 1

  Persistent, searchable semantic memory without external cloud dependencies

• 2

  That require a self-hosted, CPU-friendly memory solution with flat-file backups

• 3

  That communicate via MCP JSON-RPC protocol or HTTP calls

Description

Eidetic enables AI agents to have persistent and semantically searchable memory without relying on external cloud services. It uses PostgreSQL with the pgvector extension for efficient cosine similarity searches and Ollama for embeddings, all running locally or via Docker. The server supports a standard JSON-RPC MCP protocol, making it compatible with various agents, and offers features like flat-file backups and CPU-friendly operation on low-cost hardware.