MongoDB Query Runner for JSON

About this tool

This is a MongoDB query runner for plain JSON. Paste a JSON array of documents, paste the same find filter you would hand to db.collection.find(...), and see exactly which documents come back — without a server, a connection string, or a scratch collection. It is built for testing a filter before it ships, explaining a query to a teammate, and slicing a JSON export down to the records you actually care about.

The query language is MongoDB's query-and-projection language. Supported operators:

• Comparison — $eq, $ne, $gt, $gte, $lt, $lte, $in, $nin
• Logical — $and, $or, $nor, $not
• Element — $exists, $type ("string", "number", "int", "long", "double", "decimal", "object", "array", "bool", "null", or the BSON type code)
• Evaluation — $regex with $options, $mod
• Array — $all, $size, $elemMatch

MongoDB's matching rules are followed, not approximated: dotted paths such as team.name reach into sub-documents, a predicate on an array field matches when any element matches, and a missing field behaves the way the server behaves — {"age": null}, $ne, $nin and $not all match documents that do not have the field at all.

Worked example

Documents:

[
  {"name":"Ada","age":36,"team":{"name":"core"},"tags":["math","code"]},
  {"name":"Bo","age":24,"team":{"name":"infra"},"tags":["ops"]},
  {"name":"Cy","age":41,"team":{"name":"core"},"tags":["code","ops"]}
]

Query {"tags": {"$in": ["code"]}}, projection name, age, sort age:desc:

[
  {
    "name": "Cy",
    "age": 41
  },
  {
    "name": "Ada",
    "age": 36
  }
]

Both documents match because tags is an array and at least one element is "code". Switching the output format to Count only returns 2 instead.

Projection, sort, skip and limit

The four boxes below the filter reproduce a full cursor chain — find(query, projection).sort(...) .skip(...).limit(...):

• Projection — {"name":1,"age":1,"_id":0} in Mongo form, or the short form name, age to keep and -password to drop. _id is kept by a keep-projection unless you set it to 0, exactly as in MongoDB, and mixing kept and dropped fields (other than _id) is an error.
• Sort — {"age":-1,"name":1} or the short form age:desc, name. Values sort in MongoDB's cross-type order: missing/null first, then numbers, strings, objects, arrays, booleans.
• Skip / limit — page through a big result set. The Count only format always reports the full match count, ignoring skip and limit.

Output is a JSON array (indented by default), NDJSON for piping line by line, CSV for a spreadsheet, or just the count.

Pasting a query straight from the shell or Compass

Shell-flavoured syntax is accepted, so a filter copied out of mongosh, Compass or an application log usually runs unmodified: unquoted keys ({ age: { $gt: 21 } }), single quotes, // and /* */ comments, trailing commas, /pattern/flags regex literals, and the ObjectId(), ISODate(), NumberLong(), NumberInt() and NumberDecimal() helpers. Strict JSON is a subset, so it works too. Because the input is JSON rather than BSON, ObjectId("…") and ISODate("…") compare as the string inside them, and the numeric helpers compare as plain numbers.

Limits and edge cases

• Documents input: 5,000,000 bytes and 50,000 documents. The query, projection and sort boxes accept 20,000 bytes each, nested at most 64 levels deep.
• Everything runs locally in your browser via WebAssembly — no document ever leaves the page.
• Comparisons only relate values of the same JSON type, matching MongoDB: {"age":{"$gt":"30"}} (a string) will not match the number 36.
• $where and $expr are rejected with an explicit message — they need a JavaScript / aggregation-expression evaluator. So are $jsonSchema, $text, the geospatial operators ($near, $geoWithin, $geoIntersects), the bitwise operators, $rand and $sampleRate.
• Aggregation pipeline stages ($group, $lookup, $unwind) and update operators ($set, $inc, $push) are out of scope: this tool reads documents, it never modifies them.
• $all with $elemMatch entries and projection operators such as $slice or $ are not supported; an unknown operator always fails loudly rather than silently returning the wrong rows.
• Dates are compared as the strings they are in JSON, so ISO-8601 timestamps (2026-08-13T12:00:00Z) sort and range-compare correctly, while other date formats do not.

FAQ