chore(codegen): daily schema refresh (2026-06-18)

[timgl]

Daily OpenAPI schema refresh — 2026-06-18

Regenerated src/generated/api.d.ts via pnpm codegen from the live PostHog
OpenAPI spec (https://us.posthog.com/api/schema/?format=json), filtered by
openapi-filter.yaml. Not hand-edited.

Spec diff size

• src/generated/api.d.ts: +1783 / −137 lines
• Tracked operation set: unchanged (71 operations, identical before and after)

The drift is entirely in component schemas referenced by already-tracked
operations — PostHog evolved its dashboard tile/widget configs, experiment
configs, feature-flag metadata, and team configs. Newly pulled-in schemas
include DashboardPatchTileOpenApi, DashboardPatchWidgetOpenApi,
*WidgetConfig (Activity/ErrorTracking/SessionReplay/Experiments), several
Experiment*Config types, MinimalFeatureFlag, and Team*Config.

New operationIds added to the filter

None. Every live operationId missing from openapi-filter.yaml falls into
one of two buckets, neither of which warrants managing in the typed client:

1. Unmanaged product surfaces (the overwhelming majority) — accounts_*,
   agent_applications_*, alerts_*, annotations_*, surveys_*,
   error_tracking_*, warehouse_*, logs_*, llm_analytics_*, tasks_*,
   etc. This IaC tool manages a deliberately curated set (insights, dashboards,
   feature flags, endpoints, event definitions, property groups, experiments +
   holdouts/saved-metrics, cohorts, actions, project settings). Adding a brand
   new resource family is a product/design decision, not something an automated
   daily refresh should do unattended.

2. Auxiliary / PUT-variant endpoints on already-managed families — e.g.
   actions_bulk_update_tags_create, actions_references_list,
   cohorts_activity_retrieve, cohorts_persons_retrieve,
   event_definitions_metrics_retrieve, feature_flags_activity_retrieve,
   dashboards_collaborators_*, plus full-replace *_update (PUT) variants
   (cohorts_update, event_definitions_update, experiments_update,
   experiment_holdouts_update, experiment_saved_metrics_update) where the
   resource clients intentionally use *_partial_update (PATCH). The CRUD
   clients don't call any of these, so adding them would only bloat the
   generated types.

No new operationId represents a core CRUD operation newly available on a
managed family, so the filter is left unchanged.

Resources touched

• src/resources/dashboard/client.ts — upstream renamed the dashboard PATCH
  body schema PatchedDashboard → PatchedPatchedDashboardOpenApi. Updated the
  one type alias to match. The new schema carries the same
  restriction_level / tiles / delete_insights fields the client relies on,
  so behavior is unchanged. This was the only schema-drift breakage.

Unresolved drift

• None. No tracked operationId disappeared from the live spec (all 71 are
  still present), so there is nothing for a human to delete.
• Pre-existing lint errors in src/client/typed-posthog.ts,
  src/resources/types.ts, and src/resources/dashboard/pipeline.acceptance.test.ts
  exist on main and are unrelated to this refresh — left untouched.

Verification

• pnpm typecheck ✅
• pnpm test ✅ (291 tests, 35 files)

Originating TaskRun

Task-Id: 598ed80d-2ec6-4c5f-ac98-b680dad274aa

🤖 Generated with Claude Code