Document collections
The Database section is the workspace’s JSON document store. A collection is a
schemaless table: it has a name, a description and documents — arbitrary JSON objects.
On insert every document gets an _id field (you may supply your own value).
When you need one
Reach for a collection when the data outlives the run and you need to select it by field: orders, customer records, a catalogue, a log of messages already handled.
Other jobs have other homes. Meaning-based search over document text is a knowledge base. A single value per session — a counter, a flag, the last choice — is cheaper in variables and memory than in a collection of its own.
Database
orders
1,248 documents · 3.4 MB used
| _id | customer | status | total |
|---|---|---|---|
| a3f1… | acme | paid | 12,400 |
| b7c2… | globex | pending | 980 |
Collections are created with New collection; a document can be inserted and edited right in the interface (Insert, Edit), and the Query button runs the same filter over the collection that you will later put into a node.
The nodes
- Execute + Data
| Node | What it does | What it returns |
|---|---|---|
| db_insert | Inserts a document; an array is inserted item by item | inserted_ids, count |
| db_find | Query by filter with sorting and paging | docs, count |
| db_find_one | The first matching document | doc, found |
| db_count | How many documents match the filter | count |
| db_update | Updates by filter, supports upsert and multi | matched, modified, upserted_id |
| db_delete | Deletes by filter | deleted |
The collection is picked from a dropdown in the Collection field, not typed by hand.
Filters and updates
A filter is a subset of the MongoDB query language. You get the comparison operators
$eq, $ne, $gt, $gte, $lt, $lte; membership $in, $nin; the logical
$and, $or, $not, $nor; plus $exists, $size and $regex. Nested fields are
addressed with dots — user.name, tags.0.
An update expression understands $set, $unset, $inc, $mul, $min, $max,
$push, $pull, $addToSet.
{ "status": "pending", "total": { "$gte": 1000 } }In db_find sorting is a list of field/direction pairs
([["created_at", -1]], where 1 is ascending and -1 descending); limit defaults to
100, and there are skip and projection (1 includes a field, 0 excludes it).
Port versus field
A value arriving on the node’s input port replaces the filter (or the document) configured in the fields. When the edge delivers nothing — and an empty string from an execution-only trigger counts as nothing — the configured value stays. A value of the wrong shape fails the node with a clear error instead of turning into “match everything”.
db_update also accepts an envelope on the port,
{"filter": …, "update": …}; any other object is read as the filter.
Quota
Collection size is counted in bytes and meets two ceilings at once: the workspace’s own ceiling and the owner’s plan ceiling, which is measured across every workspace they own. Next to it sits a limit on the number of collections. Current usage is shown in the collection header (“used”) and under Settings → Usage; the limits themselves are in Plan limits.