The Complete Production SOP: How to Deploy and Run MikroORM Unit of Work Pattern & Identity Map Architecture in Autonomous AI Agents

---
name: "mikro-orm-unit-of-work-identity-map"
description: "Manage database state with enterprise Unit of Work and Identity Map patterns using MikroORM."
metadata:
  author: "MikroORM"
  version: "1.0.0"
  runtime_compatibility:
    - "codex"
    - "claude-code"
    - "cursor"
    - "windsurf"
triggers:
  - "use mikro-orm entitymanager"
  - "mikro-orm unit of work flush"
  - "identity map mikro-orm"
permissions:
  fs_read: true
  fs_write: "workspace"
  shell_exec: false
  net_access: false
---

# MikroORM Unit of Work Pattern & Identity Map Architecture

Implements MikroORM: defining entities with @Entity and Property decorators, EntityManager with fork() scoping per request, automated change tracking (flushing dirty entities), and preventing duplicate entity loads.

## Execution SOP & Rules
1. Initialize: const orm = await MikroORM.init(config).
2. Fork EntityManager per request: const em = orm.em.fork() to isolate unit-of-work transactions.
3. Mutate entities directly: user.name = 'New Name'; em.persist(newOrder); await em.flush().
4. Leverage automated dirty tracking to emit minimal SQL UPDATE statements for changed columns only.

import 'reflect-metadata';
import { defineConfig } from '@mikro-orm/postgresql';

export default defineConfig({
  clientUrl: process.env.DATABASE_URL,
  entities: ['./dist/entities/**/*.js'],
  entitiesTs: ['./src/entities/**/*.ts'],
  debug: process.env.NODE_ENV !== 'production',
});

import { Request, Response, NextFunction } from 'express';
import { RequestContext } from '@mikro-orm/core';
import { orm } from '../db';

export function mikroOrmMiddleware(req: Request, res: Response, next: NextFunction): void {
  // Creates an isolated Unit of Work and Identity Map per HTTP request
  RequestContext.create(orm.em, next);
}