Skip to content
worth noting Tools and apps

GitHub has expanded the Copilot usage metrics API to track the activity of individual third-party agents

clearly official source

GitHub has added the totals_by_3rd_party_agent field to the Copilot usage metrics API, breaking down the activity of third-party agents (Claude, Codex, etc.) by individual agent – launch and session counts for enterprise and organization reports.

GitHub has expanded the Copilot usage metrics API with a new optional field, totals_by_3rd_party_agent. The field contains one entry for each recognized agent app and is available in 1-day and 28-day reports at the enterprise, organization, enterprise-user and organization-user levels. Each entry contains agent_name (a display name that may change, which is why GitHub recommends grouping by agent_id), user_initiated_interaction_count (the number of times a user launches the agent) and session_count (the number of sessions) – session_count is available only in aggregate enterprise and organization reports and is absent from per-user records.

According to GitHub, agent activity was previously tracked in the metrics as a single aggregate indicator, making it impossible to distinguish work by Copilot coding agent from the activity of other agents, such as Claude or Codex, running through GitHub workflows. According to the company, breaking down activity by individual agent is intended to allow each agent app to be tracked separately, their adoption to be compared, and rollout and licensing decisions to be based on actual usage rather than estimates.

Access to the data requires the View Copilot Metrics permission and the Copilot usage metrics policy to be enabled. If no activity from a recognized agent app is recorded during the given period, the totals_by_3rd_party_agent field does not appear in the report at all.

What changed

Why it matters

The change is intended for administrators and teams managing GitHub Copilot deployments in companies, not for end-user developers using agents. Previously, the activity of third-party agents (Claude, Codex and others) could only be seen in aggregate in the metrics, making it impossible to determine which specific agent was actually being used and by how many people. The new detailed metrics allow Copilot license administrators to base decisions about expanding or limiting the use of a specific agent on real usage data rather than an estimate.

Relevant practical impact

What this means

01

For a business

GitHub Copilot administrators can now distinguish the activity of individual third-party agents (e.g. Claude, Codex) in the usage metrics API instead of seeing a single aggregate figure, allowing them to base decisions about rollout and licensing of specific agents on real usage data.

Processes
What to decide Enable the Copilot usage metrics policy (if disabled) and monitor the totals_by_3rd_party_agent field in enterprise/organization reports to evaluate actual adoption of individual agents before deciding on rollout and…
More business impacts →
agent apps API Copilot GitHub monitoring usage metrics

Check the original

Event sources

clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.

1
GitHub Changelog (Copilot and AI features) primary source · first detected Copilot usage metrics API adds agent app activity