GitHub published two changelog entries on October 22, timestamped 20:44 and 22:19 UTC, confirming that global default settings for Copilot features have moved to general availability. The same window brought additions to the Copilot usage metrics API, which now accounts for custom skills, agents, and MCP servers as distinct usage categories. A third, separate entry at 18:14 UTC changed how the Actions blob storage meta API endpoint behaves. None of these were framed as a single release; they landed as three independent changelog posts on the same day.
For engineers pulling Copilot usage data into warehouses or dashboards, the API expansion means new fields to map. Custom skills, agents, and MCP servers were not previously broken out as usage categories, so any existing extract-transform-load job built against the old schema needs to account for the new dimensions or risk silently dropping them into an undefined bucket. GitHub did not disclose a migration window or whether older API responses remain available in parallel, so teams should check whether the new categories appear as additive fields or replace prior aggregate counts before trusting historical comparisons.
The move to general availability for global default Copilot settings matters less for data shape and more for governance. Organizations that manage Copilot rollout through API calls or infrastructure-as-code now have a stable target instead of a preview flag that could change behavior without notice. That stability is useful for anyone scripting onboarding across multiple repositories or enterprises, since GA status typically implies a slower, more predictable deprecation path than a beta feature carries.
The blob storage meta API endpoint change for Actions, posted separately at 18:14 UTC, is the kind of update that rarely gets attention outside operator channels but can break automation quietly. Any pipeline that calls that endpoint directly, rather than through a maintained client library, should be tested against the new behavior before the next scheduled run. GitHub's changelog format does not include a diff of the exact fields or parameters affected, so the practical path is to run the endpoint against a staging workflow and compare output against logs from before October 22.
Industry comparison data now appears in the Copilot impact dashboard, letting organizations benchmark their own Copilot usage against aggregate figures. GitHub did not disclose the composition of that comparison set, including whether it draws from all Copilot customers or a subset, which limits how much weight any single team should place on a benchmark number without knowing its denominator. Taken together, the three updates reflect ordinary platform maintenance rather than a coordinated feature push, and operators who treat changelog entries as isolated events, checking each against their own integration points, will fare better than those waiting for a consolidated release note that GitHub has not indicated it plans to publish.