Product Analytics

changAIs includes optional, privacy-conscious product analytics powered by a changAIs-operated Umami instance. Collection is disabled by default and starts only after you explicitly allow it.

Product analytics is separate from the local review workflow. You can review diffs, notes, and reports without enabling it.

Consent and Controls

• changAIs asks once, after you open your first review.
• Dismissing or declining the prompt leaves analytics disabled and does not cause another prompt.
• Accepting the prompt enables the application-level changais.telemetry.enabled setting.
• You can turn collection off at any time in VS Code settings.
• VS Code’s global telemetry choice always takes precedence. If VS Code usage telemetry is disabled, changAIs sends no product analytics even when its own setting is enabled.

What We Learn

If you opt in, changAIs can record:

• the public changAIs extension version;
• that the extension or a review was opened;
• that a review was completed;
• which product features were used, such as changing note state, adding or deleting a reply, changing the diff layout, comparing report history, exporting a report, preparing an AI follow-up prompt, or running an AI-tool setup command.

These records describe the feature that was used, not the project content involved in the action.

Data We Do Not Collect

Product analytics never includes:

• source code or diff content;
• file names or workspace paths;
• Git remotes;
• report, note, reply, or user-message text;
• license keys;
• VS Code machine or user identifiers;
• error messages or stack traces;
• editor version, operating system, or device model.

The review interface does not use analytics cookies or browser fingerprinting.

Network and Location Information

Analytics events are sent to the changAIs-operated Umami service over HTTPS. Every network service necessarily sees the source IP while handling a request, but changAIs does not include an IP address in the analytics data it sends. The analytics service may derive approximate location information according to its configuration and retention policy.

Delivery is best-effort. A network or analytics-service failure never interrupts the review workflow.

For the broader privacy policy, see Privacy.