JSONBeeby Candorbees

Every JSON file is a story nobody has read yet.

The JSON viewer that reads your data for you.

Drop in a JSON or JSON Lines file and get a dashboard with KPIs, charts, search and automatic insights, with the parts that matter already highlighted. No schema, no setup, no upload.

Drop a JSON file here

or ·

Or try a sample Factory shift report LLM usage log Sensor stream (JSONL) Compare 3 shifts

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

Move the lens over the file.shift-report-PNQ02.json
I
The arrival

Open JSON, JSON Lines or a whole folder

It begins when a file lands on the page. Someone exported it from an ERP, a sensor gateway or an LLM pipeline. Nobody wrote down what the fields mean, and nobody needs to.

  • Any shape. A single JSON document or a JSON Lines stream, at any size. Whole folders too.
  • Forgiving. Trailing commas and other small mistakes don't stop it.
  • No configuration. It works out structure, types, units and meaning on its own.
  • Stays responsive. Big files are parsed in a background worker, so the page never freezes.
Dropped just now
shift-report-PNQ02.json6 machines
sensor-telemetry.jsonl120 readings
manufacturing/7 files
Three ways in. None of them asks you a question first.
II
The first reading

JSON to dashboard: KPIs, charts and tables

Then the file gets a face. Numbers turn into rupees, kilowatt-hours and minutes. Arrays become tables you can sort. The busiest field gets a chart.

  • A dashboard in seconds. A header, KPI cards, tables and charts, picked from what the data contains.
  • Real units. ₹, $, €, KB, MB, tokens, kWh, %, °C and more, shown the way people read them.
  • Drill down. Open a nested array from any row and keep going.
  • The raw tree is always there, with "copy JSONPath" on every value.
Candor Precision Components – ChakanShift A · 28 Oct 2026
Units made
2,379
Scrap
67units
OEE
71.4%
Downtime
250min
Energy
1,202kWh
Energy cost
₹10,338
Units by machine
CNC-02519
WLD-06471
CNC-01408
VMC-03386
VMC-04381
PRS-05214
$.machines[4].downtime_minutes
One press is doing half the work of the others. Keep an eye on it.
III
The questions

Search and filter JSON at any depth

Every good reader starts asking questions. Where did the time go? Which machines ran hot? You type a word and it searches every value, however deep it's nested.

  • Global search across every value at any depth.
  • Facets, ranges and dates to narrow what you see.
  • Cross-filtering. Click a bar in one chart and every table and chart follows.
  • Your call on meaning. Override how any column is read when the guess is wrong.
Searchhydraulic
hydraulic_leak · PRS-05 · 07:35$.machines[4].downtime_events[0].reason
hydraulic_leak · PRS-05 · 12:50$.machines[4].downtime_events[2].reason
running · 5 maintenance · 1
Spindle temperature
42.6 °C45 – 51 °C53.4 °C
Two leaks, both on the same press. The story is taking shape.
IV
The twist

Automatic insights that find what's wrong

Some things in the file don't add up. Rule-based insights read the whole dataset the way an auditor would. Each finding points to the exact values behind it, so you can check it yourself.

Inconsistent totalshift-report

The summary says 250 minutes of downtime. The six machines add up to 234.

$.summary.total_downtime_minutes
Duplicatework-orders

Work order WO-2026-1043 appears twice, with the same operations.

$[3] · $[14]
Round-number capwork-orders

Two orders are for exactly 1,000 units. That looks like a system default, not a real plan.

$[4].quantity · $[9].quantity
Gapsensor-telemetry

The sensors go silent for 31 minutes, from 09:39 to 10:10.

$[79].timestamp → $[80].timestamp
Outliersensor-telemetry

PRS-05 vibrates at 11.8 mm/s at 10:22. It usually runs near 3.4.

$[105].vibration_mm_s
Concentrationshift-report

One machine, PRS-05, accounts for 207 of the 234 downtime minutes.

$.machines[4].downtime_minutes

Also on the watch list: empty fields, inconsistent totals across files, and cross-file comparisons. All of it runs in your browser at no cost.

V
The ensemble

Compare JSON files side by side

One shift is a scene. Three shifts make a plot. Give it a folder and it finds the files that share a shape, stacks them, and lets you read them side by side.

  • Folder mode. Same-shaped files are stacked into one dataset, with a source column so you know which row came from where.
  • Everything still works. Filters, search, charts and insights run across all the files at once.
  • Cross-file insights. See how the headline metric spreads across files and which group pulls ahead.
  • Free. No account and no limit on the number of files.
  • No silent drops. Files with a different shape are listed as left out.
manufacturing/ · 3 shift reports stacked
ShiftUnitsDownOEE
PNQ02 · A2,379250
71.4%
PNQ02-B2,61327
78.5%
PNQ02-C2,71627
81.3%

Left out, different shape: work-orders.json, inventory-snapshot.json, quality-inspection.json, sensor-telemetry.jsonl

Shifts B and C lost 27 minutes each. Shift A lost 250, and the press is the reason.
Open this comparison
VI
A second opinion

Ask your JSON questions with AI (coming soon)

Soon you'll be able to ask the data for a second opinion. AI will be there when you ask for it and never before, and you'll see exactly what goes out before anything is sent.

  • Ask in plain words. Your question becomes a query, and the answer is computed in your browser on the full data.
  • AI insights. An interpretation, extra findings and recommendations, tied to the same JSON paths.
  • No invented evidence. Any claim that cites a path that doesn't exist is dropped.
  • A preview before you send. Redacted fields and token count, shown before anything leaves your browser.
"Which machine lost the most time on shift A, and why?"
PRS-05 lost 207 minutes, mostly to hydraulic leaks. Two hydraulic leaks and one die change · computed on all 6 machines $.machines[4].downtime_events[*]
Before you send · example preview
Redacted3 fields
Tokens~1,900
StatusComing soon
Nothing leaves your browser until you click Send.Send
∎
Epilogue

Export JSON to CSV, PNG or a report

A story is only finished once you've told it to someone. Take what you found to the shift meeting, the client or the next pull request, in light or dark.

.csvTables and chart dataFor the spreadsheet people.
.pngAny chart, as an imageFor the slide that's due in ten minutes.
.mdThe insight reportEvery finding, with its evidence, ready to share.

Your data stays in your browser. Only the answers leave.

In the browserParsing, dashboards, search, filters, insights, exports and folder comparisons, at any file size. All free, and none of it leaves your machine.
With AIComing soon, and optional. You'll preview it before anything is sent.
To startDrop a file or open a sample. No sign-up.
A
Appendix · Other viewers

How JSONBee compares to other JSON viewers

Most viewers show you the shape of a file. JSONBee also reads what's in it.

FeatureFormatters and tree viewersGraph visualisersJSONBee
Collapsible tree of the raw JSONYesYesYes
Graph of the structureNoYesNo
Dashboard with KPIs in real unitsNoNoYes
Charts picked from the dataNoNoYes
Finds problems, with the JSONPath of eachNoNoYes
JSON Lines / NDJSONVariesVariesYes
Compare many files at onceNoNoYes
Works without uploading your fileVariesVariesYes

Compared by kind of tool, not by product. Individual viewers differ, so check the one you use.

B
Appendix · Pricing

Free in your browser.

Every feature is free, at any file size, with no account. AI is coming soon.

Free

No account
  • Dashboards, search, filters, insights and exports at any file size
  • JSON and JSON Lines, with trailing commas forgiven
  • Compare several files or a whole folder
Open the app

AI

Coming soon
  • AI insights: an interpretation and recommendations, tied to JSON paths
  • Ask the data: questions in plain words, answered on the full file
  • Previewed before anything is sent
C
Appendix · Questions

JSON viewer FAQ

Is JSONBee free?
Yes. Dashboards, search, filters, insights, exports and comparing several files all work in your browser, at any file size and with no account.
Is my JSON uploaded anywhere?
No. Your file is parsed and analysed in your browser and stays on your machine. JSONBee doesn't send your data to any server.
How big a file can I open?
There's no size limit beyond your browser's memory. Parsing runs in a background worker, so the page stays responsive while a big file loads.
Can JSONBee open JSON Lines or NDJSON?
Yes. Each line becomes a row, so logs and event streams open as tables and charts. Files with trailing commas open too.
How do I compare several JSON files?
Choose several files or a whole folder in the app. Files with the same structure are stacked into one dataset with a source column, so filters, charts and insights run across all of them. It's free.
How do I find the JSONPath of a value?
The raw tree view has a "copy JSONPath" button on every value, and every automatic insight lists the JSONPaths of the values behind it.
Can I convert JSON to CSV?
Yes. Any table or chart's data exports as CSV. Charts also export as PNG, and the insights as a Markdown report.
Does JSONBee use AI?
Not yet. AI insights and "ask the data" are coming soon. The automatic insights you see today are worked out by rules in your browser, each with the JSONPaths of the values behind it.
Do I need an account?
No. Everything works without signing up.

You already have the file. Now find out what it's been trying to tell you.

Read your own file Compare the three shifts