11ai MongoDB querying

Version baseline: MongoDB 8.x, with 8.3.7 as the current stable release at this review. Inspect server, Atlas release track, featureCompatibilityVersion, mongosh, Database Tools, and driver versions independently before using 8.3-only behavior.

Turn natural-language conditions into a precise filter, projection, and bounded result plan. Confirm field names and BSON types before running a query that could return sensitive or very large data.

Filter patterns

db.orders.find({ status: "paid" })
db.orders.find({ total: { $gte: 100, $lt: 1000 } })
db.orders.find({ status: { $in: ["queued", "running"] } })
db.orders.find({ deletedAt: { $exists: false } })
db.orders.find({ $or: [{ region: "EU" }, { region: "UK" }] })
db.orders.find({ items: { $elemMatch: { sku: "A1", quantity: { $gt: 0 } } } })

Use implicit AND for conditions on separate fields. Use $or when alternatives are intended. Distinguish a missing field from an explicit null with $exists and a precise type predicate where needed.

Projection, sort, and bounds

db.users.find(
  { active: true },
  { _id: 1, email: 1, createdAt: 1 }
).sort({ createdAt: -1, _id: 1 }).limit(50)

Prefer a limit for exploratory reads. For stable pagination, use a range on a unique, ordered field such as createdAt plus _id instead of large skip values. Do not mix inclusion and exclusion projections except for _id.

Counts and distinct values

db.orders.countDocuments({ status: "paid" })
db.orders.estimatedDocumentCount()
db.orders.distinct("customerId", { status: "paid" })

Use countDocuments for an exact filtered count. Treat estimatedDocumentCount as an estimate and explain the difference.

Dates, strings, and arrays

Use BSON dates rather than date strings when the field is stored as a date:

db.events.find({
  createdAt: {
    $gte: ISODate("2026-01-01T00:00:00Z"),
    $lt: ISODate("2026-02-01T00:00:00Z")
  }
})

Use regex only when a bounded, case-aware scan is acceptable. For array membership, direct equality matches an array element; use $all or $elemMatch when multiple conditions must apply.

Verification

If a query is slow or the result is surprising, inspect actual field names and types, add a small limit, and hand off to the indexes skill for explain and index analysis. Never alter data while debugging a read query.