Ask your data · AI summaries — now shipping→

Compare CSV files.
Right in your browser.

See every changed row, resolve conflicts cell by cell, and export a clean merge — 50,000 rows in under a second. Your data never leaves your browser.

Start comparing — free

Running locally0 bytesuploadedFree · no sign-inCSV · JSON

employees_v1.csv⇄employees_v2.csv

Avatar

IDKEYAB

NameAB

EmailAB

RoleAB

DepartmentAB

Ayesha Khan

ayesha.khan@acme.coayesha.khan@acme.io

engineersenior engineer

engineering

Sara Ahmed

sara.ahmed@acme.co

designerlead designer

product

Usman Tariq

usman.tariq@acme.cousman.tariq@acme.io

engineer

engineeringplatform

John Smith

john.smith@acme.co

managerdirector

engineering

Fatima Noor

fatima.noor@acme.co

analyst

data

Omar Sheikh

omar.sheikh@acme.coomar.sheikh@acme.io

engineersenior engineer

engineering

David Miller

david.miller@acme.io

qa engineer

engineering

Layla Hassan

layla.hassan@acme.co

manager

product

AI Summary

Export resolved data

8 rows · 7 of 7 cols shown

Start here

Drop your two files.

CSV, Excel or JSON. Drag & drop, click to browse, or load sample data to explore the interface first.

100% client-side

File A

Drop file or click

.csv.json

File B

Drop file or click

.csv.json

Add two files to compare.

The toolkit

Three tools, one tab.

Everything for the CSV workflow — compare two files, view and clean a single one, or join two on a shared key. All in the browser, nothing uploaded.

COMPARE

What you get

Diff, view, and clean — in one tab.

Built for the full CSV workflow: compare two versions and ship a clean merge, or open a single file and clean it up with color rules, fill blanks, bulk replace, and column formats.

01 / Resolve3 conflicts

Accept A or B with one click.

Every conflicted cell shows both values. Pick a side or merge in bulk — every decision is reversible.

field

Ateam-q1.csv

Bteam-q2.csv

name

plan

02 / Privacy0 requests

Your data never leaves your browser.

Open DevTools and watch the network tab — not one byte of your files is ever uploaded. Parsing, diffing, and merging all happen in this tab.

NETWORK · 0 REQUESTS

> open csvdiff.app ························✓ parsed locally

> drop file_a.csv ························✓ parsed locally

> drop file_b.csv ························✓ parsed locally

> compute diff (12,481 rows) ························✓ 184ms

> resolve 23 conflicts ························✓ in memory

> export merged.csv ························✓ ~/Downloads

04 / Matchkey: id

Smart row matching.

Auto-detects keys. Renamed columns and reordered rows still line up.

06 / ExportCSV · JSON

Download merged.

CSV or JSON. Schemas, ordering, and quoting preserved.

merged_2026-04-25.csv

05 / Color rules5 rules

See patterns at a glance.

Heatmap, category, above/below average, empty cells, duplicates — five rules per column that turn walls of text into something you can scan.

employees.csv · 2 rules active

name	salary	status
Ayesha	$120,000	active
Sara	$98,000	on leave
John	$155,000	active
Zain	$87,000	pending

03 / Ask your dataruns locally

Ask questions in plain English.

Type a question, get a bar chart, pie, stat, or table — computed locally. The model only ever sees column names, never your rows.

Executes locally — row data never sent to the model

Which region has the highest average order value?

GROUP BYRegion→AGGREGATEavg(Order Value)→SORTdescending

North

West

South

East

Show North rowsAdd as column

07 / And more+6 tools

Everything else in the toolbox.

Format columns

Number, date, currency & case formats per column.

1234.5→$1,234.50

Sort, pin & filter

Reorder, pin in view, or filter to exact values.

name ↑plandate

Instant search

Fuzzy search across every cell. Press /.

ada@/

Inline edit

Click a cell to fix a typo, a header to rename.

Director

Clean with AI

Describe a fix in plain words — AI writes the rule.

="H4339"→H4339

Chart your data

Any table as a bar, donut, or line — picked for you.

08 / Joinon customer_id

Join two CSVs, not just diff them.

Match rows on a shared key and stitch two files into one — the browser alternative to a VLOOKUP or a SQL join. Inner, left, right, or outer, then filter and export.

customer_id	name
C-18	Ada

customer_id	total
C-18	$240

customer_id	name	total
C-18	Ada	$240

Join CSV files

100% client-side

Your data never leaves
this browser tab.

Every parser, diff, merge, and export runs on your device. It’s the deciding feature when the data is sensitive — and you can prove it in ten seconds.

No upload, no account

Your files are read by JavaScript in this tab. Nothing is sent to a server — there isn’t one to send to.

Safe for regulated data

PII, financial, and health records never leave your machine, so there is no data-processing agreement to sign.

Verify it yourself

Open DevTools → Network, drop your files, and watch the request count stay at zero.

NETWORK · 0 REQUESTS

> open csvdiff.app✓ served once, then offline

> drop customers_v1.csv✓ parsed locally

> drop customers_v2.csv✓ parsed locally

> compute diff (48,210 rows)✓ 212ms

> resolve 61 conflicts✓ in memory

> export merged.csv✓ ~/Downloads

// 0 bytes of your data sent anywhere.

How it works

Three steps, zero setup.

Step 1 / 03UPLOAD

a.csvb.csvdrop file

Upload both files.

Drag CSV or JSON files onto the dropzones, or click to browse. Match rows by position or by a key column.

Step 2 / 03REVIEW

−

Review every delta.

Modified, added, and removed rows are colored and counted. Drill into a row to see changed fields side-by-side.

Step 3 / 03EXPORT

merged.csvmerged.json

Export the merge.

Pick sides on conflicts (or accept all from one file), then download a clean CSV or properly-typed JSON.

Desktop app

Take it to your desktop.

Everything in the web app, plus database connections, whole folders and saved projects. Download it and try it out.

Databases, read-only
PostgreSQL, MySQL / MariaDB, SQLite and MongoDB, with SSH tunnels. Browse, filter, follow links between tables.
Whole folders
Open a folder of CSV, JSON and Excel files as one workspace: search across all of them, stack or join them.
Saved projects
Save a comparison or workspace as a .csvdiff file, with autosave and crash recovery.
Still private
Same rule as the web app: your files and rows stay on your computer.

Early accessFree · macOS, Windows, Linux

Download for macOS

Apple silicon & Intel · .dmg

Other platforms

Early builds aren’t code-signed yet, so your system may ask before opening it. macOS: if it says “Not Opened”, choose Done, then System Settings → Privacy & Security → Open Anyway (once).

All releases and notes

Blog

Guides & insights.

Practical walkthroughs on comparing CSV files online — from picking the right match key to validating data migrations and using AI to summarise what changed.

Guide8 min read

How to Diff JSON API Responses for Regression Testing — Catch Breaking Changes Before Users Do

Capture a baseline response before you deploy, capture it again after, and diff the two. A key-matched JSON diff surfaces renamed fields, dropped values, and type changes that a raw text diff buries in reordering noise — no server, no upload.

August 31, 2026Read

Guide7 min read

How to Join Two CSV Files on a Shared Key — Without Excel or SQL

Combine two CSV files into one by matching rows on a shared key column — the browser equivalent of a VLOOKUP or a SQL join. No formulas, no database, no upload. A step-by-step guide with inner, left, right, and outer joins explained.

July 26, 2026Read

Feature7 min read

Clean Messy CSV Data Online — Describe the Fix, AI Writes the Rule

Excel formula wrappers like ="H4339136200", leading apostrophes, stray currency symbols, doubled quotes, rogue whitespace — the junk that shows up in exported CSVs. csvdiff.app now lets you clean any column by describing the fix in plain English; AI turns it into a reusable find/replace rule you preview before applying. All in your browser, no upload.

July 21, 2026Read

View all 26 posts

FAQ

Common questions.

Is my data actually private?

Yes. Parsing, diffing, and exporting all run in your browser — your file contents never leave the tab. Open DevTools and watch the Network tab as you load a file; you won't see any of your data uploaded anywhere. The only outbound requests are anonymous, cookieless page-view pings (Vercel Web Analytics) and — if you've configured them — AI calls you explicitly trigger, which go directly from your browser to your provider (Google or OpenRouter) using your own API key.

What file formats are supported?

CSV and JSON. JSON arrays are flattened to a tabular view (e.g. orders.0.items.0.price) and rebuilt with proper types on export. Quoted fields, embedded commas, escaped quotes, and Windows/Unix line endings are all handled.

How big a file can I diff?

Tested comfortably up to ~50,000 rows × 70 columns on a laptop. The table is virtualized so scrolling stays smooth, but parsing and diffing happen on the main thread — very large files (multi-million rows) may briefly freeze the UI while computing.

What do the AI features actually do?

Two things, both opt-in: AI summary condenses what changed across the diff into a few bullet points, and AI match-key suggestion asks the model to recommend which column to pair rows on, with a one-sentence reason. Both send only the headers + a small sample of rows, never the full file.

Which AI providers are supported?

Google Gemini (free tier at aistudio.google.com/apikey) and OpenRouter (free models at openrouter.ai/keys). You bring your own key — we don't proxy requests or hold credits.

Where is my API key stored?

In your browser's localStorage, on this domain only. It never leaves your device except in the Authorization header of requests you explicitly trigger.

Is it free? Will it stay free?

Yes. The app is free and has no sign-in. AI features cost whatever your provider charges (most have a free tier). There's no server to operate, so there's nothing to monetize.

Compare CSV files.
Right in your browser.

Drop your two files.

Three tools, one tab.

Compare two CSVs

View a CSV

Join CSV files

Diff, view, and clean — in one tab.

Your data never leaves
this browser tab.

Three steps, zero setup.

Take it to your desktop.

Guides & insights.

How to Diff JSON API Responses for Regression Testing — Catch Breaking Changes Before Users Do

How to Join Two CSV Files on a Shared Key — Without Excel or SQL

Clean Messy CSV Data Online — Describe the Fix, AI Writes the Rule

Common questions.

Ready to diff?

Your data never leaves
this browser tab.

How to Diff JSON API Responses for Regression Testing — Catch Breaking Changes Before Users Do

How to Join Two CSV Files on a Shared Key — Without Excel or SQL

Clean Messy CSV Data Online — Describe the Fix, AI Writes the Rule

Ready to diff?

Compare CSV files.Right in your browser.

Drop your two files.

Three tools, one tab.

Compare two CSVs

View a CSV

Join CSV files

Diff, view, and clean — in one tab.

Your data never leavesthis browser tab.

Three steps, zero setup.

Take it to your desktop.

Guides & insights.

How to Diff JSON API Responses for Regression Testing — Catch Breaking Changes Before Users Do

How to Join Two CSV Files on a Shared Key — Without Excel or SQL

Clean Messy CSV Data Online — Describe the Fix, AI Writes the Rule

Common questions.

Ready to diff?

Your data never leavesthis browser tab.

How to Diff JSON API Responses for Regression Testing — Catch Breaking Changes Before Users Do

How to Join Two CSV Files on a Shared Key — Without Excel or SQL

Clean Messy CSV Data Online — Describe the Fix, AI Writes the Rule

Ready to diff?

Compare CSV files.
Right in your browser.

Your data never leaves
this browser tab.

Your data never leaves
this browser tab.