Data and privacy

At Unleash, we prioritize the privacy and security of our customers’ data. Our architecture ensures privacy by evaluating feature flags locally within the client SDKs or Unleash Edge, meaning no user data is shared with the Unleash instance.

Unleash collects anonymous usage counts to help measure usage statistics and improve the product. The following sections explain what data we collect and how you can manage data collection settings.

What data is collected

When running Unleash, we collect the following data:

Version and instance ID

We collect a unique identifier and version number for your Unleash instance. This ID allows us to measure usage statistics and the adoption of Unleash across different installations. It also helps us ensure that you’re using the latest version with the most up-to-date features and security enhancements.

Feature usage data

v5.3

To improve Unleash, we collect anonymous usage statistics:

• Number of active feature flags
• Number of projects, environments, users, user groups, and custom roles
• Number of active segments, custom strategies, and custom context fields
• Number of times features flags have been exported or imported

All collected data is anonymous and limited to usage counts. This data helps us understand how features are used in Unleash, enabling us to prioritize important features and make informed decisions about deprecating features that are no longer relevant to our users.

MCP server feedback

MCP v0.5.0 Beta

The Unleash MCP server can share anonymous diagnostic feedback reports with Unleash to help improve the MCP server. The assistant sends a report when a tool call fails, when a result isn’t what it expected, or when you ask for something the MCP server can’t do yet.

• Local MCP server: Feedback reports are opt-in. The MCP server sends reports only after you accept a one-time consent prompt, and saves whether you opted in or out on your machine.
• Remote MCP server: In Unleash v8.3 and later, an admin turns feedback reports on or off for the whole instance. See Feedback from the remote server.

Each report contains only the following fields:

• Issue type: tool_error, unsupported_action, or unexpected_result
• MCP tool name and a normalized error code, such as HTTP_500
• MCP server version, and the name and version of the MCP client
• A short summary that the assistant writes

The assistant is instructed to leave flag names, project IDs, code, URLs, and tokens out of the summary. Reports never include raw error messages or your personal access token. By default, reports go to Unleash, not to your Unleash instance. You can also send reports to your own endpoint. For details, see Send feedback.

Manage data collection settings

We recognize that privacy requirements differ across organizations. While the data collected by Unleash is limited and anonymous, we provide options to manage data collection settings:

Disable all telemetry

Set the CHECK_VERSION environment variable to anything other than true, t, or 1. This disables both version and feature telemetry.

Disable feature telemetry

Set the SEND_TELEMETRY environment variable to anything other than true, t, or 1 before starting Unleash. This disables feature telemetry while still allowing version tracking.

Disable MCP server feedback

For the local MCP server, set UNLEASH_MCP_SEND_FEEDBACK=false in the MCP server’s environment. You don’t see a consent prompt, and the assistant doesn’t see the send_feedback tool. If you distribute MCP configuration centrally, set the variable there to apply it to every developer.

For the remote MCP server, go to Admin settings > Remote MCP server and disable Send feedback from the Remote MCP Server to Unleash.