PRME Developer Documentation

Complete guide to integrating PRME (Portable Relational Memory Engine) into your application.

Guides

Guide	Description
Getting Started	Installation, quickstart, and your first memory-powered app
Core Concepts	Node types, lifecycle states, scopes, epistemic model
MemoryClient API	Synchronous Python SDK (recommended for most use cases)
Async Engine API	Low-level async API for advanced use
CLI Reference	Command-line tool for setup, inspection, and maintenance
MCP Server	Model Context Protocol server for Claude, Cursor, and other MCP clients
HTTP API	REST API for language-agnostic integration
Configuration	Environment variables, scoring weights, and tuning
Deployment	PostgreSQL backend, encryption at rest, production setup

Architecture Overview


Your Application
      |
      v
 MemoryClient (sync) or MemoryEngine (async)
      |
      v
 +-----------+-----------+-----------+
 | Ingestion | Retrieval | Organizer |
 | Pipeline  | Pipeline  | Jobs      |
 +-----------+-----------+-----------+
      |
      v
 +---------+---------+---------+
 | DuckDB  | usearch | Tantivy |
 | Events  | HNSW    | FTS     |
 | + Graph | Vectors | Search  |
 +---------+---------+---------+
   Optional: PostgreSQL backend

Quick Links

Store a memory: client.store()
Search memories: client.retrieve()
LLM extraction: client.ingest()
Run as MCP server: prme-mcp --db-path ./memories
Run as HTTP API: python -m prme.api

Source: documentation/README.md in PRME at 58caddb1.