GitHub Copilot reports now name the custom skills, agents and MCP servers in use
GitHub's Copilot usage data now shows which custom tools teams actually use, so admins can see what is worth keeping and what is idle.
The short version
- The October 22 update swaps generic labels like "other" and "custom" for the configured names of each customization.
- Covered types: skills, custom agents, MCP servers, slash commands and plugins.
- Each category now lists up to 25 entries per record, up from five; the rest falls under "other."
- Older reports are not updated with names, and outdated or blocked client telemetry can cause undercounting.
- Viewing the data needs the right enterprise or organization permissions and an enabled usage metrics policy.
Why it matters
Custom agents and MCP servers take effort to build, maintain and secure. Until now, admins could say who used Copilot but struggled to say which of their own add-ons earned their keep. Named usage data lets them retire unused tools and back the ones people actually reach for.
What GitHub changed
As Visual Studio Magazine reported, GitHub expanded the API behind its Copilot usage metrics. An API is simply a way for software to request data, and admins use this one to pull reports on how their people use Copilot.
Before the October 22 update, activity from add-ons often showed up under broad buckets such as "other" or "custom." Now the reports carry the names that teams gave to their own skills, custom agents, MCP servers, slash commands and plugins.
Reading the new reports
MCP, short for Model Context Protocol, is a standard way to connect an AI assistant to outside tools and data. Slash commands are typed shortcuts that trigger a specific action.
The names appear in enterprise and organization reports, including the ones broken out by individual user and the combined totals. Every category can now list 25 entries per record, where the old limit was five. Anything beyond that is still pooled under "other."
The underlying counts of interactions and distinct uses did not change. Only the labeling got more precise.
Limits to know about
The detail works across the Copilot command-line tool, the Copilot app, clients built on the Copilot SDK (a developer toolkit) such as Visual Studio Code, and integrations using the Agent Client Protocol. Some clients need version 1.0.97 of the command-line tool or 1.0.21 of the SDK, or something newer.
GitHub warned that numbers can come in low if the software is out of date or if telemetry, the usage data clients send back, is blocked. Past reports will not be filled in with names after the fact.
What admins can do with it
The practical gain is a clearer picture of internal tools. A team that built a custom agent can now see whether anyone opens it, and a team weighing a new MCP server can point to evidence of demand for the ones it already has.
Access requires the right permissions at the enterprise or organization level, plus the Copilot usage metrics policy switched on.
What to watch
- Will the 25-entry cap be enough for organizations with large libraries of custom tools?
- How much will blocked or outdated telemetry skew the numbers in real deployments?
Try it
If you administer Copilot, confirm the usage metrics policy is on and your team's tools are updated, then pull a fresh report to see which custom tools are named.
Sources and links
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by Visual Studio Magazine · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.