Understand any JSON payload in one glance

The first thing you do with an unfamiliar API response is figure out its shape. Which fields exist? Which are nullable? Is items an array of objects or an array of arrays? The docs might tell you — if they're current — but the payload in front of you is the ground truth.

QDev reads that ground truth for you.

Schema view: the shape, not the sample

Switch any open document to the Schema tab and QDev infers a JSON Schema from the actual data — every key it saw, the type of each value, which fields are optional across records, and how deep the nesting goes. It renders in the same interactive tree as the data, so you can collapse the branches you don't care about and drill into the ones you do.

On a single record that's convenient. On an array of ten thousand, it's the only sane way to see the shape: QDev merges what it observes across all of them, so a field that's present in most records but missing in a few shows up as optional — exactly the edge case that breaks a naive parser.

From "what's in here" to "give me just this"

Once you can see the shape, the ⌘K bar turns it into extraction. Because the AI "ask" mode is grounded in that inferred schema, plain-English questions resolve to correct paths on the first try:

the id and email of every user without a verified_at

You get the jq, the result, and a match count. If you'd rather drive it yourself, type jq directly — it runs locally in WebAssembly, no server round-trip.

Big files don't change the workflow

Schema inference is most valuable precisely when the file is too big to eyeball, so it can't fall over on size. QDev streams large documents and keeps the tree virtualized, so inferring the schema of a multi-hundred-megabyte export works the same as a toy example — the tab stays responsive.

The point

You shouldn't have to reverse-engineer a payload by scrolling it. Let the data describe itself, then ask it for exactly the slice you need.

Read the docs → · QDev for data engineers →