GitHub Fixes Missing Copilot Agent Activity in Usage Metrics
An IDE-attribution gap left some agent use out of enterprise reports or assigned it to Copilot CLI, and the missing historical data cannot be recovered.
Edited by Tyronne Panaino
GitHub says an attribution defect caused some Copilot agent activity to disappear from usage reports or appear as Copilot CLI activity after several IDEs moved agent sessions to the Copilot SDK. The October 6 changelog notice says the affected sessions did not identify their originating IDE, preventing GitHub's reporting system from assigning them correctly.
The fix is available in Visual Studio Code 1.139.0 and later. GitHub expects updates for Visual Studio and JetBrains IDEs during October, followed by Eclipse and Xcode by November. The repair is forward-looking: developers have to move to a fixed client before new agent activity is counted correctly, and GitHub says the missing historical data cannot be backfilled.
Why reported agent activity fell
The defect affected only IDE versions that use the Copilot SDK for agent mode. GitHub says earlier versions remained counted. For affected clients, most agent interactions were left out of reports because the sessions lacked an IDE identifier, while some activity was attributed to Copilot CLI instead.
That produces two different distortions. Agent activity and agent-edited lines of code can appear lower than they were, while the Copilot CLI category can appear higher. GitHub says earlier CLI inflation cannot be corrected because the wrongly attributed events cannot be separated from genuine CLI use after the fact.
The notice applies to enterprise, organization and user reports in both one-day and 28-day views. It also distinguishes reporting from billing: GitHub says charges were not affected. The issue changed how activity was attributed in analytics, not what customers were billed.
Client updates repair future data, not history
VS Code is the only client listed as fixed immediately, beginning with version 1.139.0. GitHub names Visual Studio 18.12 as expected in October and the next JetBrains plugin release as expected by late October. The next Eclipse and Xcode plugin releases are expected by November. Those dates are rollout targets, not evidence that every listed client is already fixed.
Organizations that manage IDE versions centrally face a practical reporting boundary. Users who remain on an affected SDK-based version continue to produce incomplete agent metrics, while users on a fixed release resume correct attribution. Because the older records cannot be reconstructed, trend lines should recover gradually rather than jump to a retrospectively corrected history.
The telemetry design explains the gap
GitHub's Copilot usage-metrics documentation says most detailed metrics come from client-side IDE telemetry. Server-side signals can identify active users when client telemetry is missing, but richer breakdowns such as feature use and lines of code remain empty until the client supplies the necessary data.
That split explains why overall Copilot use could keep growing while agent-specific detail fell. A server can observe that a user made a request, but it does not automatically see what happened inside the editor or which client produced every detailed event. Network controls, disabled telemetry and outdated extensions can create similar reporting gaps, so installing a fixed IDE is necessary for this defect but may not resolve every analytics discrepancy.
What administrators should watch
Teams using Copilot metrics for adoption reporting should record client versions alongside the date each fixed release reaches their fleet. Comparisons spanning the affected period need an explicit caveat: missing agent detail and inflated CLI attribution are known data-quality limits, not proof that developers reduced agent use.
The next verifiable checkpoint is completion of GitHub's client rollout through November. GitHub has not quantified how many users or events were affected, the size of the undercount, or the exact release dates for the non-VS Code fixes. Those limits prevent a universal correction factor from being inferred.
Evidence quality and status
Confirmed. GitHub's changelog documents the defect, affected reporting, available VS Code fix, planned client releases and non-backfillable history. Its documentation explains the client/server telemetry boundary. Internal confidence is medium because both sources are controlled by GitHub and no independent audit quantifies the reporting error.
Sources
Update note: Last reviewed 2026-10-07. We will revise this post as fixed IDE versions ship or if GitHub publishes quantified impact or revised reporting guidance.
Sources
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.