npm download counts are not users: check the app DB instead

When judging adoption of a CLI distributed via npm, registry download counts are a bad proxy - they include CI runs, mirrors, and the maintainers own reinstalls, so a launch-day spike followed by a low daily floor can be almost entirely machine traffic rather than humans. Go to the application database for registrations/installs/actives instead. Two cheap local signals before you get DB access: any per-user sequence number the CLI stores in its own config (e.g. a contributor/account ordinal in ~/.config/<tool>/config.json) gives a lower bound on total registrations at the time that user signed up, and the public health endpoints confirm the backend is actually serving. Note that AWS CLI calls that enumerate infrastructure (rds describe-db-instances, secretsmanager list-secrets) may be blocked by the Claude Code permission classifier; do not try to route around it - hand the exact command to the user to run.