GitHub Says Copilot Usage Reports Undercounted Agent Activity From Some IDEs

GitHub has said that its Copilot usage metrics undercounted agent activity. It did so in a changelog post on its own blog, Update your IDE to restore agent activity in Copilot usage metrics, dated October 6, 2026. The post says the missing data cannot be filled in afterward. Parts of the post follow, not the whole.

What the Reports Showed

The post is addressed to customers whose Copilot usage metrics have shown agent activity, or agent lines of code, falling while Copilot usage kept growing. GitHub says it has found the cause, and a fix is rolling out to each IDE.

Several IDEs recently moved Copilot agent sessions to the Copilot SDK, the post says. Those sessions did not identify which IDE they came from. So the usage metrics could not attribute them correctly. Most of that activity was left out of reports, and some was counted as Copilot CLI activity.

Only IDE versions that use the Copilot SDK for agent mode are affected, the post says. Developers on earlier versions are still counted.

The reports kept arriving. Nothing in a falling line says whether the work fell or the counting did.

What to Expect in the Reports

At the time of the post, the fix was available in one IDE. Other IDEs will receive it in upcoming releases, which GitHub expects to finish rolling out by November 2026. Once developers update, their agent activity is counted again in the Copilot usage metrics dashboard and API.

The gap persists for any developer on an affected IDE version until they update, the post says. Their agent interactions and agent lines of code stay undercounted. GitHub says this applies to enterprise, organization and user reports, for both 1-day and 28-day reports.

GitHub says it "can't backfill missing data." Activity from affected IDE versions does not identify which IDE it came from, so it cannot be attributed after the fact. Customers should expect a gradual recovery as developers move to the fixed versions, rather than a single jump. And developers on an affected version should update as soon as the fixed version is available, the post says, because their activity cannot be recovered later.

Copilot CLI metrics may be inflated, the post adds, because some activity from other SDK-based clients was counted as CLI activity. This will clear up as developers update those clients. But earlier CLI metrics cannot be corrected, because that activity cannot be separated from real CLI use. Billing is not affected, GitHub says: the issue changed only how agent activity was attributed in usage metrics, not what customers were charged.

Where the Detailed Numbers Come From

Most detailed usage metrics come from telemetry that each IDE sends, the post says. GitHub also records server-side data when Copilot handles a request. That data reliably shows who was active, but it cannot see what happens in the editor.

When reports and other Copilot data disagree, the post says, the gap is usually on the client side. Its list of causes includes telemetry turned off in the IDE, and a proxy or firewall that blocks the telemetry endpoint. It also includes a client that changed how it sends telemetry, as with this change.

GitHub says it is steadily reducing how much reports depend on client telemetry. Usage metrics now use server-side data to count active users that client telemetry misses, and to identify the IDE for those users. But some detail, such as lines of code and accepted suggestions, can only come from the editor. So client-side gaps won't disappear entirely, the post says.