# Ogham MCP

An MCP server that gives AI coding agents a shared, persistent memory backed by Postgres, so context carries over between sessions and between different agent clients.

Ogham MCP is a persistent, searchable memory server for AI coding agents that works across clients such as Claude Code, Cursor, and OpenCode. Rather than losing context every time an agent session ends or a developer switches tools, agents write what matters to a shared store and retrieve it again later through Ogham's retrieval engine.

Retrieval combines hybrid search with a typed-edge graph rather than vector similarity alone, which lets it answer questions whose answer is a path between two stored facts, a case plain retrieval-augmented generation handles poorly. A setup wizard connects a Postgres database, such as a free Supabase project or a Neon database, picks an embedding provider, migrates the schema, and writes the MCP client configuration for supported clients automatically.

It installs with uvx, Docker, or from source, and requires a Postgres database with the pgvector extension.

Website: https://ogham-mcp.dev