Claude Code keeps only your own history, so a team view needs everyone's. Usage Insights for Claude Code gets there without a server and without anyone's prompts: each person exports counts and names, and one person loads the files.
Everyone exports their events
Each person opens their own folder, then presses Export in the bar. They type a pseudonymous id (made up, never an email or a name) and the team id everyone agreed on, such as team_platform, and choose Events (the dialog also has the full reports and two other data files).
The file is saved to their own disk. It holds one line per event in the catalog's unified schema: times, counts, costs, and the names of models, tools, skills and project folders. No prompt, no answer and no session title. If a folder name says more than you want to share, export one project at a time with the project filter.
Share the files inside your team
Collect the files the way your team already shares files. They never pass through BENIBIX: the page has no server, and nothing in it can send data anywhere.
Load them all at once
The team lead opens Usage Insights for Claude Code, presses Choose files and picks every .jsonl file at once. The Team view fills in:
- DAU / WAU / MAU and weekly active people against your seats; type the seat count to see unused seats
- Stickiness: average daily active people as a share of monthly active people
- Depth of use: light, regular and power users, where power means agents, skills or connectors
- Outputs, estimated time saved as a range, and spend per active person
Retention and adoption by team
New & returning users shows who started each week and who came back after a month away. Adoption by team shows the share of each team that is active. Retention cohorts follows the people who started in the same week, and the share still active each week after.
Small groups stay hidden
A team with fewer than five people is merged into Other teams, or hidden when even that is under five, and the panel says so. Cohorts under five people are left out. Nobody can be picked out of an aggregate, and there are no named rankings.
Use cases, skills and connectors
Further down, Top use cases compares topics per team, Org skills & plugins ranks the skills most people reuse, Connector coverage shows which apps each team uses, and Tokens by model and Spend show consumption over time.