[P1] infra/dashboards/ is empty but operations.md claims four committed JSONs

[audriB]

Files:

• docs/operations.md:104-110 (§Dashboards)
• infra/dashboards/ (directory, empty)

Issue

operations.md §Dashboards states:

Committed as code under infra/dashboards/:

• overview.json
• cloud.json
• auth.json
• business.json

The directory exists but is empty.

Impact

During an incident, the responder follows the ops doc, runs ls infra/dashboards/, finds nothing, loses firefighting minutes figuring out that Prometheus metrics exist but no ready-made dashboards do.

Fix

Two paths:

(a) Commit the four JSON dashboards — export from Grafana as JSON model, stamp into infra/dashboards/. Then document the re-import flow.

(b) Delete the §Dashboards section of operations.md. Users can build dashboards ad-hoc against the documented metrics.

(a) is clearly better long-term (incident response benefits from curated views) but (b) is faster and stops the live lie.

Related

• Related: HIGH-D3 (rollout-health.yml referenced, doesn't exist) — same kind of drift
• Audit doc: docs/reviews/Audit_2026-04-23.md (HIGH-D4)

[audriB]

Partial progress in #76: operations.md §Dashboards no longer claims four committed JSONs under infra/dashboards/ — instead it points at the actual Prometheus metric names for ad-hoc Grafana dashboards. Exporting the dashboards themselves to JSON + committing them to the directory is still open; tracked here until done.