# kairix/worker.py — RATCHETED OFF (SGO-109, branch agent/sgo109-worker-coverage).
# The _default_* production-default seams (_default_embed, _default_health_check,
# _default_neo4j_drain, _default_wal_checkpoint, _default_connector_sync,
# _default_entity_seed, _default_neo4j_client_for_seed) are now executed at unit
# scope via their public callers with deps=None (the PR #605 WarmStackDeps /
# Neo4jDrainTickDeps pattern — see tests/test_worker_default_seams.py), asserting
# real degraded behaviour in the unit env. worker.py now measures 94.52% unit
# (F7) — comfortably above the 90% floor with a ~4.5-point buffer that resolves
# the historical 90.1-90.6% variance concern — so it no longer needs grandfathering.
# kairix/connectors/m365_calendar/graph_client.py — Graph SDK boundary
# adapter for calendar events. Uncovered branches are Graph-specific
# error envelopes whose contract belongs to F68 + integration tests,
# not F7.
kairix/connectors/m365_calendar/graph_client.py
# kairix/worker_cli.py — CLI dispatcher; the uncovered lines are
# argparse subcommand wrappers that call into already-tested worker.py
# entry points and immediately exit. F30 (operator-outcome tests)
# covers the subcommand surface end-to-end via subprocess invocation;
# F7 doesn't catch additional defects here. Currently at 90.1%, kept
# on baseline because 1 line of variance flips CI per the worker.py
# rationale; the right paydown is the same _default_* wrapper extraction
# pattern, tracked in followup work.
kairix/worker_cli.py
