Skip to content
UsageAtlasBeta
TodayHistoryInsightsAlertsPrivacy
GitHubOpen source
UsageAtlas home

Privacy, mapped

Privacy policy

Use UsageAtlas without an account, or create a free account to save usage history. This policy explains local data, optional cloud storage, analytics, and your choices.

Effective September 9, 2026

On this page

01Overview02Data the desktop app processes03How that data is used04Optional accounts and cloud saves05Anonymous usage count06Network requests and third parties07Website data08Storage and retention09Your choices10Security11Changes to this policy

01

Overview

UsageAtlas is a local-first desktop application for viewing AI-provider capacity and usage. This policy explains what the desktop application and this website process.

You can use the desktop app without a UsageAtlas account. Usage is processed and stored on your computer. If you want to keep a cloud copy and restore saved history after changing or losing a computer, you can create a free account with Google or GitHub.

Cloud saving is optional. Signing in from the desktop turns on hourly saving by default, unless you previously turned it off on that computer. The anonymous install count is separate from cloud save and can be switched off in Settings.

02

Data the desktop app processes

Depending on the providers you enable, the desktop app may process:

  • provider identity and plan information;
  • quota windows, remaining capacity, credits, and reset times;
  • token counts, request counts, model names, timestamps, project and session labels;
  • estimated costs calculated from bundled pricing data; and
  • local preferences and redacted diagnostics.

For Codex, UsageAtlas uses the signed-in CLI's local app-server API for account capacity and reads local usage logs for activity. For Claude, it uses supported local credentials and usage logs. For Cursor, it can read the signed-in desktop session from a local read-only database and request bounded usage history. For OpenCode, it reads the local OpenCode database and, where available, clearly labelled local quota estimates.

03

How that data is used

UsageAtlas uses provider and local-activity information to authenticate supported provider requests, calculate usage aggregates, render the dashboard, remember settings, and diagnose errors on your device. If you sign in for cloud save, usage summaries are also stored under your account so you can restore them later. The app separately checks for desktop updates.

The renderer receives bounded dashboard aggregates. It does not receive provider credentials, raw event rows, database rows, log contents, prompts, or response bodies. The anonymous usage count never includes provider credentials, provider identities, prompts, response content, file paths, project names, usage totals, costs, or quota values.

04

Optional accounts and cloud saves

Creating an account and saving usage history are free. An account is not required for local tracking. If you choose cloud save, Google or GitHub handles sign-in, and UsageAtlas stores account information in PlanetScale Postgres. This may include your name, email, profile image, sign-in provider identifiers and tokens, session records, and technical information such as IP address and browser details used for authentication and security.

Signing in from the desktop enables automatic saving by default, unless you previously turned it off for that account on this computer. Pending changes are saved once an hour while the app is running. You can turn off Save automatically in Settings and use Save to cloud manually. Restore from cloud retrieves saved history for the same account.

Cloud saves include daily token and request totals, estimated costs, hourly activity, public model breakdowns, dates, time zones, collection status, and the latest limits and plan information. Source and record identifiers let the app match records across saves. These records are linked to your account; they are not anonymous analytics.

Cloud saves exclude project and session names, file paths, prompts, responses, raw logs, and the credentials used by your coding tools. Those credentials are separate from the Google or GitHub sign-in tokens used for your UsageAtlas account. Restoring a cloud save restores usage summaries, not every local project or session detail.

Signing out of the desktop stops future saves but does not delete previously saved cloud history. You can request deletion of your account and cloud history at privacy@usageatlas.com. The desktop protects its saved account session with your operating system's encryption service.

05

Anonymous usage count

UsageAtlas keeps a general count of how many people use the app. The desktop app registers the installation and a few basic events, such as opening the app or installing an update, along with the app version and operating system. That is all the anonymous count sends.

This count does not include your UsageAtlas account ID, name, or email. It also excludes prompts, provider details, usage totals, and costs. A random identifier stands in for the installation, and the forwarded IP address is replaced at our proxy. You can turn the desktop count off in Settings whether or not you use cloud save. Turning off this count does not turn off cloud saving; each has its own setting. Events are processed on UsageAtlas's behalf by PostHog through a Cloudflare proxy that removes cookies.

06

Network requests and third parties

When you refresh a supported provider, the desktop app sends requests directly to that provider's service using the provider session already available on your device. Those requests are governed by the provider's own privacy terms.

Packaged Windows and macOS builds periodically contact the Cloudflare-hosted UsageAtlas update service. The website, downloads, and release files are also delivered by Cloudflare. Standard technical request data may be processed by Cloudflare to deliver, protect, and observe those services. Cloudflare also handles account and cloud-save requests, and PlanetScale stores account information and saved usage records. Google and GitHub process sign-in through their own services and privacy terms. Anonymous product events are processed by PostHog.

07

Website data

You can browse the website and download the app without an account. Public pages use PostHog to measure visits and interactions, including download clicks. PostHog may store a random browser identifier in browser storage or cookies. Account pages are excluded from this analytics capture.

If you choose to sign in, the website uses session cookies to keep you signed in and process desktop connection requests. These account cookies are separate from analytics. Google and GitHub may also process data when you use their sign-in pages.

Cloudflare may process ordinary delivery and security logs, including IP addresses, request times, requested pages, and browser information. The edge service also records aggregate download and update-request counters without provider or dashboard data.

08

Storage and retention

Desktop preferences and usage history are stored in the local application data directory until you remove that data. Uninstalling the app may leave this directory behind. Local history lets you view previously collected usage when a provider is unavailable.

Account information and cloud usage records are kept so you can sign back in and restore saved history. Signing out, turning off automatic saving, or uninstalling the app does not delete existing cloud records. Request account and cloud-history deletion at privacy@usageatlas.com. Only history successfully saved to the cloud can be restored; cloud save does not copy your entire local database.

Anonymous product events and infrastructure logs are retained only as long as reasonably needed for product measurement, delivery, security, abuse prevention, and legal obligations. Retention may vary between Cloudflare and PostHog.

09

Your choices

You can use UsageAtlas without an account, disable individual providers, and turn the anonymous usage count on or off. If you choose a free account, you can turn automatic saving off, save or restore manually, or sign out of the desktop to stop future saves. You can also remove local app or browser data, uninstall the app, or request deletion of your account and cloud history.

Depending on where you live, you may have rights over personal data, including access, correction, deletion, restriction, or objection. Requests can be sent to privacy@usageatlas.com. You may also contact your local data-protection authority.

10

Security

UsageAtlas uses a sandboxed renderer, context isolation, narrow typed operations, navigation restrictions, bounded provider responses, diagnostic redaction, a separate utility process for provider access, HTTPS security headers, and streamed private release storage. No system can be guaranteed completely secure, but the app is designed to reduce the amount and exposure of sensitive data.

11

Changes to this policy

This policy may change as UsageAtlas evolves. Material changes will be reflected by a new effective date on this page. Review the current policy when you install a new release or use a new website feature.

UsageAtlasBeta

One private picture of all your AI coding work.

Product

TodayInsightsSourcesDownload

Guides

All guidesClaude Code historyCheck Codex usageOpenCode usage

Legal

Privacy policyTerms of usellms.txtRSSGitHub

© 2026 UsageAtlas contributors.

Codex, Claude, Cursor, and OpenCode are trademarks of their respective owners.