MongoDB

Pro-tier engine. The full Mongo experience — collections, BSON editor, aggregation pipeline builder, change streams, native dump.

Connection

Field	Notes
URI	mongodb://user:pass@host:port or mongodb+srv://…
Database	Default database for the session
TLS	Auto-detected from mongodb+srv://; manual override available
Auth source	Optional — defaults to the DB in the URI

For Atlas: paste the SRV URI from the Atlas dashboard.

Native idiom — the document mindset

Quay treats Mongo as Mongo, not as “a SQL database with JSON columns”. The browser surfaces:

• Database → Collection → Documents as the rail hierarchy
• BSON editor for per-document edits — type-aware (ObjectId, Date, Decimal128, Binary, etc.) with type-preserving inline edits
• Schema sample — collection-level shape is inferred from the first 100 docs because Mongo is schemaless. The “schema” tab shows field paths, sampled types, and per-field null/missing rates.

Edit-as-draft for documents

Same buffer mechanics: edit a field in the BSON editor, the change goes pending. Multiple field edits across documents stage together; ⌘S commits them in a single bulkWrite operation (atomic per collection).

Aggregation pipeline builder

MongoBrowser → Aggregation pipeline runner (Pro tier deep panel). Stage-by-stage UI:

• Each stage is an operator ($match, $group, $lookup, …) + its body as JSON
• Reorder stages by drag-and-drop
• Run produces a result grid (or another collection if $out / $merge is the last stage)
• Save the pipeline as a snippet for re-use

Change streams tail

watch() cursor on the cluster, database, or single collection. Live tail panel:

• Filter by operation type (insert / update / delete / replace)
• Pretty-printed fullDocument per event
• Click an event to jump to the affected doc in the browser

Native dump (mongo_dump_database)

The native backup walks every collection, writes a // ===== <name> ===== marker per collection, then emits one JSONL line per document. The file is restoreable through Quay’s import (which understands the marker format), and is also valid input to mongorestore if you’d rather use the upstream tool.

The format is .jsonl, not BSON, so it’s human-readable + grep-able. The trade-off: BSON-specific types (Decimal128, Binary) round-trip through the standard MongoDB extended-JSON format ({ "$numberDecimal": "1.234" }).

Edge cases worth knowing

• _id editing — Mongo doesn’t allow updating _id on an existing doc; Quay surfaces this as a soft warning when you try (the buffer accepts the edit; the commit step warns + skips it with an explicit acknowledgement).
• Regex search — find/grep across docs goes through Mongo’s $regex operator. For large collections, prefer $text (text index) via the search box’s “Use $text” toggle.
• Replica sets — Quay opens a single primary connection; for read-from-secondary use the URI’s readPreference=secondary.

What’s NOT yet wired

• Atlas Search index management — surfaced read-only; create / drop is via the Atlas UI for now (write API support planned).
• Field-level encryption (CSFLE) — Quay decrypts client-side if your --keyVaultNamespace is reachable; otherwise encrypted fields show as <encrypted>.