Files
zeekay ec6f6b50c1 chore: kill ALL posthog — zero refs, zero filenames (1452 -> 0)
We own the ingestion server (hanzoai/insights: Django + rust/capture + nodejs
plugin) AND this SDK, so there is no external wire contract to preserve. The
server is already 100% posthog-free; this SDK was the last laggard.

- $last_posthog_reset -> $last_insights_reset. This FIXES A LIVE MISMATCH: the
  server taxonomy (insights/taxonomy/taxonomy.py, insights/insightsql/ai.py)
  already expected the insights form while the SDK emitted the posthog form.
- __PosthogExtensions__ -> __InsightsExtensions__; all PostHog/posthog/POSTHOG
  identifiers -> Insights/insights/INSIGHTS (src, tests, playground, playwright,
  scripts, rrweb subpackages, package.json keywords).
- Hosts: DEFAULT_PLUGIN_HOST + DEFAULT_NUXT_HOST + the core host default pointed
  at PostHog's cloud (us.i.posthog.com) — a real bug. All -> insights.hanzo.ai.
- Removed the us-assets/eu-assets CDN split in getRemoteConfig (PostHog-specific;
  Hanzo serves remote config from the same host) — it had degenerated to a no-op.
- Fixed generate-docs writing to a stray references/@hanzo/ subdir.
- Renamed the 3 posthog-named test files; dropped 58MB of stale reference
  snapshots named for upstream posthog versions we never shipped (they are
  generated output — references/ is now gitignored).

Verified: typecheck errors 42 (HEAD) -> 18 (all pre-existing test-mock
looseness); types package 0; every package.json valid with @hanzo/* names.
Guard: grep -rioE 'posthog' packages ... | wc -l  => 0
2026-07-16 09:09:18 -07:00
..