v0.13.9

Added

• New read-only MCP tool alertint_usage_stats reports an operational usage
  summary over a time window: alert deliveries and alerts received, LLM
  call/token volume with a per-model breakdown, Slack cards posted (new
  incident cards only) and skipped, incident analyses completed and triage
  exhaustions. Every counter is derived from the audit log, so historical
  windows stay stable as new deliveries arrive. Slack notify.sent audit
  rows now carry new_card so a resolved card posted for a never-notified
  incident counts as a card; migration 0013 adds a composite (kind, ts)
  audit-log index so windowed reads walk only the window.

Changed

• Docs: new Kubernetes (Helm) page
  covering the Artifact Hub listing (verified publisher), signature
  verification, the values schema and a production values layout; the
  README and Quickstart point at it.

Fixed

• Resolved alert deliveries now revisit their existing incident memberships
  before any collecting-window or group-key fallback. An older incident can
  therefore reach resolved even when a newer incident shares its group key;
  the transition checks every member atomically and only its winner publishes
  the recovery notification. (#97)
• Coordinated releases now keep Helm README badge labels, chart tests, and
  release-script fixtures aligned with the current versions, so the release
  metadata commit passes the normal main CI checks.

---

Full Changelog: chart-v0.2.1...v0.13.9