Fill feature-footprint gaps found preparing a customer account-review slide

Description

Preparing a customer Grafana Cloud footprint slide (one tile per product: Scaled / Adopting / Opportunity) from the published views, several tiles could not be filled or needed hand-aggregation across views. Each gap below should land as a point-in-time view row (zero new series) unless a trend genuinely justifies a metric under budget.py rules.

Gaps:

1. Agent Observability (AI Observability): nothing collected. doc-0006 lists grafanacloud_org_agent_o11y_generations_included_usage (org billing); need per-stack adoption (stacks sending generations, generation volume over a stated window) from a verified source.
2. Database Observability: doc-0006 lists grafanacloud_instance_active_dbo11y_* but it is not in coverage_capability_adoption. Add a row (stacks with active dbo11y instances, instance count).
3. Application Observability and Kubernetes / Infra Observability: no adoption row or host-hour figure; the slide sourced host hours elsewhere. Add rows from the app_o11y / infra_o11y host-count and host-hour families with the window stated.
4. Adaptive Telemetry as one capability: coverage_capability_adoption has no Adaptive Metrics / Logs / Traces rows. Stacks with applied rules, pending recommendations and removable volume had to be summed by hand from cost, cost_adaptive_logs, cost_adaptive_metric_recommendations and coverage_adaptive_traces_inventory. Also cost_summary publishes null for ‘Adaptive rules applied’ and the dependent rows while cost.json carries per-stack applied counts on 309 of 310 stacks; confirm whether that null is the intended unknown-propagation (GCI-0067) and, if so, surface the measured-subset figure with its coverage instead of a bare null.
5. Enterprise plugins: usage_plugin_adoption and usage_datasource_inventory do not classify Enterprise datasource plugins, so the count was done against a hand-typed list. Classify from a live source (plugin catalogue signature/plan metadata), never a literal list, and add an Enterprise rollup (stacks, configured datasources by plugin). Query-level usage depth is only available on the few stacks usage-insights covers; state that coverage.
6. Assistant users: ai_summary reports a sum of per-stack active users; the org figure (grafanacloud_org_assistant_users) would de-duplicate and match what account teams quote.
7. Mobile Observability (private preview): record whether any signal exists yet; park if none.

Customer-facing figures that prompted this live in the consumer engagement, not here; do not copy any customer identifier into this repo.

Acceptance Criteria

• #1 Each gap 1-7 is either implemented as a view row with its population basis and window, or recorded in doc-0006 as unobservable with the route or scope that blocks it
• #2 Agent Observability per-stack adoption is sourced from a verified route or metric, compared against an authorised control before a zero is trusted
• #3 coverage_capability_adoption includes Adaptive Metrics, Adaptive Logs, Adaptive Traces, Database Observability and Application Observability rows, and the cost_summary null for applied rules is either fixed or documented as intended with the measured-subset figure shown alongside
• #4 Enterprise plugin classification comes from live plugin metadata, with no literal plugin list in code
• #5 No new metric is added without a budget.py CATALOGUE entry and a stated trend or alert justification

Definition of Done

• #1 just test
• #2 just tf-validate
• #3 just check-identifiers and just no-em-dashes both return clean

Implementation Plan

loop20: implement only gaps 2,3,4,6 in frozen owned files; qualifying existing observations; gate and independent review before landing; live proof remains R-dev10.

Implementation Notes

Loop12 children .01-.05 accepted with qualified zero-product-series views and explicit unavailable gaps in the matrix. Parent remains partial: named robk Agent positive-generation control not observed, exact cumulative generation volume and host-hour contracts unresolved; DB reporting markers do not establish units/adoption, Assistant org source empty and default lookback unknown. Do not mark parent AC2 complete or infer zero activity. Library source/publication accepted default-off with no customer grant. Follow-up scope/attempt authorization belongs to a later owner-graded packet.

loop20 bounded gaps2,3,4,6 accepted b5f703ed1f1e126677e833fe9ca8a2cd9b940fa1: dbo11y/App marker populations and separate org App/Infra host-hour-named observations state 24h query observation, unverified units and producer window; no database-instance/adoption or billable-hour claim. Missing families absent/unknown. Existing Adaptive rows/applied-rule subset and org Assistant beside sum preserved. Independent review PASS exact patch621ff623; CodeRabbit minor numeric precision fixed. Composed just check exit0 2587 passed,2 existing skips,tofu32; exact CI37951780833 all jobs success. Zero product series; VIEW_INPUTS re-derived unchanged. Parent stays In Progress: excluded gaps1,5,7 and live R-dev10 evidence not claimed; do not mark whole parent criteria done. loop20:1 implementation attempt,0 review-repair rounds; bounded code accepted.

loop21: 0 implementation attempts; parent closure withheld because R-dev11 (dev v0.11.0 rollout) is blocked on cross-repository preflight refusing unrelated dirty lock file. No dev readback performed; D-FOOT21 qualification remains authorized but acceptance prerequisite unmet.

loop22: 0 implementation attempts, 0 review-repair rounds for parent closure. Closed under D-FOOT21 after accepted R-dev22 (dev v0.11.0 rollout and readback); all six feature-footprint subtasks remain Done. Parent AC2 stays unchecked: named robk Agent positive control is still unobserved, so no verified zero/adoption or exact cumulative generation volume is claimed. Historical generic DoD boxes unchanged per goal. Qualified live evidence is retained under /Users/rob/repos/grafana-cloud-org-insights/codex/loop22-evidence/R-dev22/.

Final Summary

loop22 qualified closure under D-FOOT21, not a completion claim for unresolved product semantics. AC1: gaps 1-7 map to Done children and doc-0006 (feature-usage observability matrix); unavailable cumulative Agent volume, Mobile sessions/use, installed-version Enterprise entitlement and semantic/unit gaps remain documented. AC2: NOT VERIFIED; robk Agent observation is unknown, not zero; live per-stack rate view has 2 positive measured and 4 unknown of 6 live stacks over 30d, not generation counts or sustained use. AC3: immutable dev T2 2026-10-09T19:08:47+00:00 has Adaptive Metrics/Logs/Traces, App and DB rows; applied rules is a qualified 0 on 5 of 6 measured unsegmented stacks, never an estate total. DB markers and App host markers each report 3 of 6 live stacks with usage/opportunity null; database-instance and host units/adoption remain unverified. Org App/Infra host-hour-named families are unavailable; 24h query observation is not a verified producer accumulation window or billable host-hour unit. AC4: live current-public-catalogue Enterprise rollup has 7 plugin rows, 2 Enterprise and 2 unclassified, not installed-version entitlement or activity. AC5: feature-footprint delivery adds no product metrics; prior catalogue/gate evidence from accepted subtasks is retained. Assistant org gauge is unavailable beside per-stack rolling-30d sum 3, not a deduplicated org user figure. Six fresh live stacks discovered. Exact one dev T2 arn:aws:ecs:eu-west-1:515966537543:task/rkps-robk-insights/16c62c78e741436fafe9bf491e56d171 exited 0 on verified v0.11.0 consumer; immutable publication and 12-dashboard readback, exact schedule restoration and no-change plan verified. Evidence: /Users/rob/repos/grafana-cloud-org-insights/codex/loop22-evidence/R-dev22/runtime-readback-summary.json and /Users/rob/repos/grafana-cloud-org-insights/codex/loop22-evidence/R-dev22/footprint-live-summary.json.