Customization activity
Track reported usage, inspect adoption, and find workflows worth a closer look.
Activity over time
Activity by category
Worth a closer look
Data-backed suggestions, not productivity scores.
Daily adoption
Intensity is relative within each row. An unlisted name is not proof of zero usage: names can fall outside a report's cap. Select a name to inspect its history. How adoption is measured →
Customization inventory
| Days observed |
|---|
Data & coverage
Loaded reports
Metric definitions and interpretation
Reported activity counts skill and slash-command invocations, custom-agent starts, and MCP connection attempts. These are activity signals, not unique actions, time saved, or measures of individual performance.
Plugins overlap skills. Plugin counts are excluded from combined activity totals, but plugins remain visible in the inventory and can be explored as a separate category. MCP counts are connection attempts, not tool calls.
Names and reach are lower bounds. Reports retain up to 25 named entries per record; other/custom buckets, unavailable telemetry, and reporting caps can hide activity. Inventory includes both GitHub-provided and customer-defined names; a real report does not include an ownership flag.
Reach means observed users in your uploaded reports, not all licensed developers. Per-user period totals only support reach for their exact full window. Aggregate activity takes precedence when matching aggregate and per-user files are loaded; they are never added together.
Recent change compares the latest seven days with the previous seven, or shorter equal periods for shorter selections. Missing days or changes in reporting coverage prevent unsupported comparisons. A newly observed name is not necessarily a newly launched customization.