Privacy policy

finopsway-costs browser extension. Last updated 27 September 2026.

This is the privacy policy for the finopsway-costs extension, as published in the Chrome Web Store and on Firefox Add-ons. It describes what the released extension does.

The short version

The extension has no backend. The only network requests it makes go to the Databricks workspace you are signed in to. We run no server, receive no data and cannot see what you query.

What the extension handles

A Databricks personal access token, if you choose token mode. It is stored with chrome.storage.local, keyed by workspace origin. That store is readable by this extension alone, is never synced to a browser account, and is cleared when you remove the token from the options page or uninstall the extension. In session mode no credential is stored: the extension reuses the workspace session you already have, and the CSRF token it needs is fetched per request and held in memory for the duration of that request.

Your workspace account name, in session mode only. So that a query runs as you rather than as nobody, the extension asks the workspace who you are: it reads /auth/session/info and, where that answers with a numeric id alone, /api/2.0/preview/scim/v2/Me. The name it gets back (an email address) is sent to your own workspace as the identity the statement runs as. It is held in memory for that request, is never stored, and is never sent anywhere but the workspace it came from. Token mode looks nothing up: a personal access token is already an identity.

Databricks billing usage records and SKU prices. The extension runs read-only SQL queries against the system.billing tables of your own workspace and keeps the rows in IndexedDB, in the workspace origin's own storage. This is a cache: it holds the history you configure on the options page (Days to keep, a year by default, which is everything Databricks itself retains, and never more), is pruned to that retention on every sync, and can be dropped at any time from the options page. The most recent five weeks are kept at the hourly grain the table has; anything older is folded to one row per day per SKU and compute resource, so the cache holds less detail about a day the further back it is, never more. A row carries the cost figures and the ids of whatever ran up the cost - workspace, cluster, warehouse, job and run, pipeline, app, serverless compute - and, for a job or a Databricks App, the name Databricks records for it, which is what lets the toolbar name what you are looking at instead of printing an id. Alongside the rows it keeps only its own bookkeeping: which day was last fetched, when, and how many rows landed. It contains only records the signed-in identity is already entitled to read.

Compute machine hours, if you use the cloud cost view. To estimate what the machines under your clusters cost, the extension also reads system.compute.node_timeline - again read-only, again from your own workspace. It selects only the cluster id, the machine type, whether the node was a driver, the hour, and counts of minutes and instances. It does not read that table's utilisation, network or IP address columns. The result is aggregated to one row per cluster, machine type and hour before it is stored, and is replaced wholesale on each refresh rather than accumulated.

Which cloud and region your workspace runs in. The extension runs SELECT current_metastore(), which returns a string such as aws:us-east-2:<id>. The cloud and the region select which machine price list to use; the metastore identifier in that string is stored alongside them but is not read by any feature, is never written to the console, and like everything else here never leaves your browser.

Extension settings. Your warehouse choice, catalog, retention, display preferences, which cost figure the toolbar shows, whether each page integration is on, whether debug logging is on, and the period the toolbar is showing (a preset such as "month to date", or the dates you chose and what they are compared against), in chrome.storage.local.

A little per-workspace bookkeeping, on the same store and keyed by the same origin: the workspace's warehouse list as the options page last read it (each one's id, name and running state), a record of the query currently in flight so another tab can pick it up instead of asking the warehouse twice, when the last sync finished and how many rows it returned, and a flag saying the extension is active on this workspace. Removing the workspace on the options page, or uninstalling, removes all of it.

What the console page in front of you shows, where you switch a page integration on. The cost badges and the SQL warehouse hourly price are drawn inside Databricks' own pages, so they read those pages: the links a list row carries (which is where the cluster, job, pipeline, app or warehouse id comes from), the size and type a warehouse's row or form is set to, and the "N DBU / h" rate the console prints. That reading happens in your browser and goes nowhere; the badges are added to the page and nothing of the console's own content is edited. Each integration has its own switch on the options page, and there is a master switch for all of them.

Where data goes

Every network request the extension makes is to the Databricks workspace host you are visiting, matching *.cloud.databricks.com or *.azuredatabricks.net. There are no other hosts, and the extension declares no permission that would allow any. Your token is sent to that workspace and nowhere else. Credentials are keyed by workspace origin, so a token for one workspace is not reachable from another.

The cloud cost view contacts no cloud provider. AWS and Azure machine list prices are downloaded by us in advance, from those providers' public price lists, and ship inside the extension package as static files. Estimating what a cluster costs is a lookup in a file already on your disk. Nothing about your clusters, your account or your region is sent to Amazon, Microsoft or anyone else, and the extension has no permission that would let it.

What we do not do

• We do not collect, receive, store or transmit any of your data. There is no server to send it to.
• We do not sell or transfer your data to third parties.
• We do not use your data for anything unrelated to the extension's single purpose, which is displaying your Databricks spend.
• We do not use your data to determine creditworthiness or for lending.
• There is no analytics, telemetry, crash reporting or advertising of any kind.
• The extension writes nothing to Databricks. Its SQL only reads system.billing and system.compute, and only ever with SELECT.
• It does not read the content of your work. No notebook, query text, table, result or file is read at any point. The tables it queries describe cost and compute usage, not your data, and what the page integrations read off the page is the console's own labels and links, never anything of yours.

Logging

Debug logging is off by default, in released builds as in any other, and while it is off the extension writes no diagnostic traces at all. You can turn it on from the options page (Debug logging) when something needs diagnosing; leave it off otherwise, because the extension logs to the page console, which the workspace app can read and anyone looking over your shoulder can see. Even with it on, the access token and the CSRF token are never printed, and automated tests enforce that. What the trace does name is the authentication mode, your settings, the workspace id and, in session mode, the account the query runs as.

Removing your data

Uninstalling the extension removes the stored token, the settings and the bookkeeping above; Remove this workspace on the options page does the same for one workspace. The cached billing rows, machine hours and the cloud/region record live in the workspace origin's IndexedDB and are removed together by the Clear cached usage action on the options page, or by clearing site data for the workspace in your browser.

A note on token scope

A Databricks personal access token has the same permissions as your user account. It cannot be limited to system.billing alone. The extension only uses it for read-only billing queries, but you should treat it like any other workspace credential: prefer a short-lived token and revoke it in Databricks when you no longer need it. Session mode does not store a credential at all.

This website

finopsway.com counts page views with a self-hosted GoatCounter instance. It sets no cookies and stores no personal data: only the page, the referrer, the browser and screen size, and the country derived from the IP address, which is not stored. This applies to the website only, never to the extension.

Trademarks

finopsway-costs is an independent extension built by FinOpsWay. It is not affiliated with, endorsed or sponsored by Databricks, Inc.

Databricks is a trademark of Databricks, Inc. Google Chrome and the Chrome Web Store are trademarks of Google LLC. Firefox and addons.mozilla.org are trademarks of the Mozilla Foundation. Microsoft Edge and Azure are trademarks of Microsoft Corporation. AWS is a trademark of Amazon.com, Inc. or its affiliates. These names are used only to describe what the extension works with. All trademarks remain the property of their respective owners.

Contact

Questions about this policy: mail@finopsway.com.