Telemetry: measure Apso usage (opt-out anonymous CLI telemetry + passive npm/PyPI stats)

[cultron]

Summary

Add a way to measure how many people use Apso (usage numbers / trends). Separate from the commit co-author hook — this is its own initiative with trust/privacy implications.

Signals

A. Passive (free, available now — do first)

• npm download stats for @apso/cli and @apso/domain-events.
• PyPI download stats (pypistats) for apso-domain-events.
• Coarse (downloads ≠ active users; CI inflates) but zero work and a usable trend.
• Go module proxy has no easy public per-module counts.

B. Active — opt-out anonymous CLI telemetry (the real signal)

The CLI emits an anonymous event per command run. No PII, no code, no schema contents.

• Events: command name (generate/init/deploy/…), CLI version, OS/arch, an anonymous install id (random uuid persisted in CLI config), coarse counts only (e.g. number of entities — optional, decide).
• Transport: an analytics sink — PostHog / Segment / a small first-party ingest endpoint (decide).
• Privacy/trust (required):
  • Opt-out via APSO_TELEMETRY=0 and a persisted config flag (apso config set telemetry off).
  • First-run transparency notice ("Apso collects anonymous usage data; opt out with …").
  • Respect DO_NOT_TRACK.
  • Documented in README + docs; never collect identifiers or source.

C. Platform-side (already partial)

apso deploy / login / link hit the Apso backend, so authenticated platform usage is already loggable server-side — but not local-only generate.

Proposed sequencing

1. Stand up the passive npm/PyPI dashboards now (no code).
2. Design + implement opt-out anonymous CLI telemetry (B): pick the provider, add the client + anonymous id + opt-out + first-run notice + docs, stand up the ingest.

Out of scope

The Co-Authored-By: Apso commit hook is a separate feature (attribution, not telemetry; only visible on public-repo commits).

Environment

• @apso/cli