Memgoose

Use this skill to work efficiently with memgoose using only the documented API. memgoose is a lightweight MongoDB-like, Mongoose-style TypeScript database with memory storage by default and optional persistent storage.

First Steps

Identify the task and load only the needed reference:

• Testing or mocking mongoose: read references/testing.md.
• Basic models, CRUD, and connection setup: read references/basics.md.
• Schema fields, validation, indexes, timestamps, methods, or statics: read references/schemas.md.
• Queries, updates, chaining, select, sort, pagination, lean, populate, or atomic operations: read references/queries.md.
• Aggregation pipelines and analytics: read references/aggregation.md.
• Storage backend choice or persistence setup: read references/storage.md.
• Performance tuning: read references/performance.md.

Prefer docs-backed patterns over assumptions from Mongoose. If a requested Mongoose feature is not in these references, inspect the project docs before claiming support.

Common Workflows

Use memgoose in Tests

Mock mongoose before importing app code or production models:

import { vi } from 'vitest'
import { connect, disconnect, dropDatabase } from 'memgoose'

vi.mock('mongoose', () => import('memgoose'))

beforeAll(() => connect({ storage: 'memory' }))
afterAll(async () => disconnect())
beforeEach(async () => dropDatabase())

For route tests, register the mock before building the app and then exercise the app through its test interface. See references/testing.md.

Build a Model

Use Schema and model for default memory storage, or create a configured database with connect() / createDatabase() when storage matters.

import { Schema, model } from 'memgoose'

const userSchema = new Schema({
  email: { type: String, required: true, unique: true },
  age: { type: Number, min: 0 }
})

userSchema.index('email')

const User = model('User', userSchema)

See references/basics.md and references/schemas.md.

Query and Update Data

Use MongoDB-like filters and update operators. Query builders are thenable and also support .exec().

const users = await User.find({ age: { $gte: 18 } })
  .sort('-age name')
  .limit(20)
  .select('email age')
  .lean()
  .exec()

await User.updateOne({ email }, { $inc: { loginCount: 1 } })

See references/queries.md.

Choose Storage

Use memory for tests and transient data. Use file, SQLite, or WiredTiger only when persistence is required or the task explicitly targets those backends. Always disconnect persistent databases when done.

See references/storage.md.

Guardrails

• Do not describe memgoose as full Mongoose or MongoDB parity; use the documented supported APIs.
• Do not add undocumented configuration keys or operators.
• Use indexes for frequently queried fields and compound indexes for common multi-field filters.
• Use lean queries only when virtuals, methods, and save() are unnecessary.