JSONBeeby Candorbees

JSON Lines and NDJSON viewer

A JSONL viewer where every line becomes a row.

Drop in a log, an event stream or a sensor feed and get a table, charts over time, search and automatic insights. No schema, no import step, no upload.

Drop a JSONL or NDJSON file here

or ·

Or try a sample Sensor stream (JSONL)

Free at any file size · No sign-up · Your data stays in your browser

sensor-telemetry.jsonl · 120 lines
{"timestamp": 1793244600, "sensor_id": "VIB-11", "machine": "CNC-01", "vibration_mm_s": 2.49, …}
{"timestamp": 1793244600, "sensor_id": "VIB-12", "machine": "PRS-05", "vibration_mm_s": 3.49, …}
{"timestamp": 1793244660, "sensor_id": "VIB-11", "machine": "CNC-01", "vibration_mm_s": 2.14, …}
…117 more lines
What you see first
Readings
120
Peak vibration
11.8mm/s
Alarms
1
Ninety minutes of readings from two machines. The press averages 3.5 mm/s and peaks at 11.8: that's the line to look at.
I
Reading the stream

Open JSON Lines and NDJSON without converting them

Most JSON viewers expect one document and stop at line two. A JSONL file is a stream, so JSONBee reads it as one: every line is a row.

  • Any extension. .jsonl, .ndjson, or a .json file that turns out to hold one object per line. Blank lines are skipped.
  • Nested lines stay nested. Arrays inside a line open as child tables, and the tree view shows the JSONPath of any value.
  • Timestamps become time. ISO dates and Unix timestamps in seconds or milliseconds are read as dates, so charts run over time.
  • Big logs stay responsive. Parsing and profiling run in a background worker, off the page.
  • Several files at once. Daily logs with the same shape stack into one table, with a column saying which file each row came from.
II
What it notices

Find the gaps, spikes and duplicates in your logs

A stream of a thousand lines hides the three that matter. The automatic insights point at them, with the JSONPath of every value behind each finding.

  • Gaps in time. A sensor or a service that went quiet for longer than usual.
  • Outliers. Readings, latencies or token counts far from the rest.
  • Duplicates. The same event logged twice.
  • Empty fields. Fields that are null on half the lines, often a device or a version that never sends them.
  • Concentration. One user, machine or endpoint behind most of the volume.
III
What people open

Logs, events, sensors and LLM outputs

If a program writes it one line at a time, it opens here.

  • Application logs. Structured logs from pino, bunyan, zap, structlog or a log shipper's export.
  • Analytics events. Product events exported as NDJSON, one event per line.
  • IoT and sensor data. Readings from gateways and data loggers.
  • LLM batches and datasets. Batch API results, evaluation outputs and fine-tuning files, with token counts as KPIs.
  • Database exports. mongoexport and other tools that write one record per line.
?
Questions

JSONL viewer FAQ

What is a JSONL file?
JSON Lines, also called NDJSON (newline-delimited JSON), is a text file with one JSON value on each line. Logs, event streams, sensor readings, API exports and LLM batch outputs often use it, because a program can append one line at a time.
How do I open a JSONL file?
Drop it on this page or choose it in the app. Each line becomes a row in a table, and the dashboard, charts and insights are built from those rows. It works the same for .jsonl, .ndjson, and a .json file that turns out to hold one object per line.
Is my file uploaded anywhere?
No. The file is parsed and analysed in your browser and stays on your machine. Parsing runs in a background worker, so the page stays responsive while a large log loads.
What about nested objects inside each line?
Nested objects and arrays stay nested. The table shows each line's top-level fields, arrays open as child tables, and search reaches every value at any depth. The raw tree view shows the exact JSONPath of any value.
Can I convert JSONL to CSV?
Yes. The table of rows exports as CSV, with any filters you applied. Charts export as PNG and the insights as a Markdown report.
Is it free?
Yes, at any file size and with no account.

Every line was written for a reason. Find the ones that matter.

Open your own file Try the sensor stream