Home › Guides › Working offline

Using SQL Offline and on Air-Gapped Machines

Once the page has loaded, the network is no longer needed. You can disconnect entirely and keep opening files, running queries and exporting results. Here is exactly how far that goes, including the one part that degrades.

Try it right now

This takes thirty seconds and is more convincing than anything written here:

Open the tool and let the page finish loading.
Turn off your Wi-Fi, or unplug the cable. Genuinely disconnect.
Now drop a file in, write a query, run it, and export the results.

Everything still works. Nothing is fetched and nothing is sent, because the parsing and the SQL engine are already in your browser.

Why this is possible

The whole application — the SQL engine, the CSV and Excel parsers, the WebAssembly modules that read Parquet and Avro — is delivered when the page loads and then runs locally. There is no backend to call. A tool that had to send your file somewhere could not survive step 2 above, which is what makes the test worth doing.

Load it once, then disconnect and keep working.

Open the tool →

Keeping a copy for repeated offline use

Staying on one tab forever is impractical. To keep a local copy:

Open the tool while connected.
Press Ctrl+S (Cmd+S on a Mac) and choose Web Page, Complete.
You get an HTML file and a folder of supporting files. Keep them together.
Open that HTML file any time, with or without a connection.

The one thing that degrades

⚠ The SQL editor uses Monaco, which is loaded from a CDN. With no connection and no cached copy, it cannot load — so the editor falls back to a plain text area. You lose syntax highlighting and line numbers. Everything else is unaffected: queries run normally, Ctrl+Enter still works, all file formats still parse, and every export still works. It is a cosmetic downgrade, not a functional one.

This is worth knowing in advance rather than discovering on a plane. If a highlighted editor matters for your offline use, the internally hosted build bundles it so there is no CDN dependency at all.

Genuinely air-gapped machines

For a machine with no network at all:

On a connected machine, save the page as above, or obtain the self-contained build.
Transfer the folder on approved removable media, following your own media-control procedure.
Open the HTML file in a modern browser on the air-gapped machine.

It runs from local disk. There is nothing to install, no service to start and no admin rights required — usually the deciding factor on a locked-down build.

For a supported deployment

If this is for an organisation rather than a laptop, a proper internal build is a better fit than a browser save: a self-contained folder to host on your own intranet, with every dependency bundled, plus a signed statement that it makes no network requests and SHA-256 checksums for every file so your security team can confirm what they deployed is what they reviewed. Available on request.

What still works offline

Capability	Offline
CSV, TSV, TXT parsing	Yes
Excel (.xlsx, .xls, .xlsm)	Yes
JSON, XML, YAML	Yes
Parquet and Avro (WebAssembly)	Yes
All SQL: joins, aggregates, filters	Yes
Export to CSV, JSON, Excel, Parquet	Yes
Syntax-highlighted editor	Falls back to a plain text area

Where this actually matters

Flights and trains. The classic case — prepare before you lose signal.
Secure facilities where the machines have no route out by design.
Restricted corporate networks that block file uploads outright.
Incident response on a network deliberately isolated.
Field work with unreliable connectivity.

In every one of those, the alternative is usually “wait until you are back online”.

OmniSelect FileSQL