GitHub CLI introduces default telemetry collection

GitHub CLI collects telemetry data with version 2.91.0. The data collection by GitHub's official command-line tool is client-side, pseudonymized, and active by default. Users who do not want this can opt out.

According to GitHub, the data collection serves to better understand the agentic use of GitHub CLI. The development team needs insight into the practical use of the functions. The collected data would land in GitHub's internal analysis infrastructure. After the telemetry feature was initially only mentioned in the release notes for version 2.91.0 and included in the documentation, GitHub has now also officially communicated it in its blog.

The documentation for GitHub CLI names more than 20 fields such as agent, architecture, is_tty, and skill_names that are the focus of telemetry. A logging mode can be activated via the environment variable export GH_TELEMETRY=log or the configuration option gh config set telemetry log. If logging is active, the tool provides insight into the telemetry data collected, but without actually transmitting it.

Three ways to opt out

Users can, in principle, prohibit GitHub CLI from collecting data in three ways: by setting the environment variable export GH_TELEMETRY=false, where a false-equivalent value like 0 or disabled also works; furthermore, via export DO_NOT_TRACK=true; and thirdly, via CLI configuration with gh config set telemetry disabled.

GitHub already began analyzing user data in April of this year. A change in the terms of service has allowed the Microsoft-owned software development platform to use user interactions with Copilot Free, Pro, and Pro+ for its AI model training. Business and enterprise accounts are excluded. Here too, there is an option to opt out. Worth mentioning: The Arch Linux package disables GitHub telemetry by default from version 2.91.0-3. For other distributions, users must take action themselves.

(mro)