DataCharter blog

DataCharter on your desktop

August 2, 2026 · Rishi Mashelkar

Confession: I love the terminal. uvx datacharter serve feels like home. But every time I showed DataCharter to an analyst friend, the demo died at the same spot — not at contracts, not at masking, at “first, open a terminal.”

So: DataCharter is now a desktop app.

The DataCharter desktop app

Double-click, pick a workspace folder (or start with the demo), and you’re in — the explorer, guides, evals, the audit timeline, the governed agent surface, all of it. The app remembers your last workspace and keeps a Workspace ▸ Recents menu, like the document-based apps you already know.

Same engine, zero compromise

There is no “desktop edition.” The app is the exact DataCharter you install from PyPI — the local FastAPI server and web UI — wrapped in your operating system’s own webview (WKWebView on macOS, WebView2 on Windows). It binds to localhost only. Every governance layer applies identically: PII masking, row filters, the read-only guard, the flight recorder, canary tripwires. Your data still never leaves your machine; it just stopped requiring a terminal to prove it.

Under the hood it’s Python frozen with PyInstaller and a pywebview window — about 90 MB, cold-starts in a couple of seconds, and each platform’s build runs a headless smoke check in CI before it’s allowed near a release.

The honest part: it’s a beta

Downloads live on the latest release — macOS .dmg (Apple Silicon + Intel) and a Windows .exe.

Two things to know:

The terminal paths — brew install datacharter/tap/datacharter, uvx datacharter — remain the first-class citizens. The app is for the days (and the colleagues) that don’t want one.

Download → open → pick a folder → your data, governed.

Code and docs: github.com/datacharter/datacharter · Desktop app docs · datacharter.dev

Try DataCharter in 60 seconds. One local process, no config, no account. Your data stays on your machine.

$ uvx datacharter serve

Star it on GitHub · datacharter.dev