v0.2.0

Added

• Continuous integration (.github/workflows/ci.yml) — runs on every push and pull
  request to main: parses every script under Windows PowerShell 5.1 (the toolkit's
  compatibility target), scans for PowerShell 7-only syntax, and runs PSScriptAnalyzer
  against a checked-in PSScriptAnalyzerSettings.psd1. The build fails on any parse error
  or Error-severity finding; Warnings are surfaced as annotations for triage.
• Subscription / RG structural-limit visibility (Get-SubscriptionLimits.ps1) — a read-only
  collector for ARM control-plane limits that silently block deployments but are not capacity
  quotas. It counts live objects via Azure Resource Graph and the ARM control plane and compares
  each against the documented limit on Microsoft Learn: resource groups per subscription (vs 980),
  tags applied to the subscription (vs 50), and resources per resource group per type (vs 800,
  near/at-limit rows only). Opt-in switches add tags-per-resource density (-IncludeTagDensity,
  vs 50), subscription deployment history per location and distinct deployment locations
  (-IncludeDeploymentHistory, vs 800 / 10), role assignments at/below the subscription scope
  (-IncludeRoleAssignments, vs the fixed 4000), and an Informational resources-per-region
  inventory (-IncludeRegionInventory). Every row carries LimitSource=MicrosoftLearnDocumented,
  IsTrueQuota=False and a DocReference to the exact Learn article — these are documented ARM
  constants, not adjustable capacity quotas, and they do not feed the compute-only quota-groups
  feature. Reader-only. Resolves #16.
• PaaS quota visibility (Get-PaasQuota.ps1) — a read-only collector for the data-PaaS tier.
  Azure SQL is reported as true quota from the Microsoft.Sql usage APIs: subscription/region
  counters (ServerQuota, VCoreQuota, RegionalVCoreQuotaForSQLDBAndDW, Managed Instance vCore
  quotas, etc.) and per-logical-server DTU quota (server_dtu_quota), each emitted generically with
  used / limit / available / PctUsed and OK / NearLimit / AtLimit flags; free-tier /
  promotional countdown counters (*Free*, *DaysLeft) are classified Informational so a healthy
  "full" countdown is not mistaken for exhaustion. Cosmos DB has no verified subscription/region
  RU/s quota API, so it is reported honestly as inventory: the per-account configured
  totalThroughputLimit (a real limit only when positive) and, optionally
  (-IncludeCosmosThroughputInventory), provisioned RU/s per SQL-API database and container, all
  flagged IsInformational. Scope via -Service AzureSQL|CosmosDB|All; supports -Location,
  -AllLocations and the standard subscription selectors. Does not feed the compute-only
  quota-groups feature. Reader-only. Resolves #15.
• Storage quota visibility (Get-StorageQuota.ps1) — a read-only collector that reports the
  per-subscription, per-region storage-account count quota (the data behind
  az storage account show-usage, default limit 250) with used / limit / available / PctUsed and an
  OK / NearLimit / AtLimit flag (threshold via -NearLimitPct, default 80). Optionally
  (-IncludeDiskCapacityInventory) it adds an informational managed-disk capacity inventory —
  total provisioned GiB by region and disk SKU, summed from Resource Graph — explicitly flagged
  IsQuota=False / Informational with a blank limit, because Azure exposes no general
  per-subscription managed-disk capacity quota usage API. Supports -Location, -AllLocations and the
  standard subscription selectors. Reader-only. Resolves #14.
• App Service quota visibility (Get-AppServiceQuota.ps1) — a read-only collector that joins App
  Service Plan inventory (Resource Graph) with the subscription/region
  Microsoft.Web/locations/{loc}/usages and per-plan Microsoft.Web/serverfarms/{name}/usages APIs
  (api-version 2025-05-01). Emits three row scopes — SubscriptionRegion, AppServicePlan, and an
  InventoryDerived row comparing each plan's current instance count to its documented tier scale-out
  ceiling (Basic 3 / Standard 10 / Premium v1 20 / Premium v2-v4 30 / Isolated 100) — with
  used / limit / available / PctUsed, NearLimit / AtLimit and PlanAtInstanceCeiling flags.
  True API-reported quota rows (IsTrueQuota=True) are kept distinct from documented/inventory rows via
  LimitBasis; SKUs without a fixed ceiling (Consumption / Flex Consumption) are flagged
  HasUnknownLimit rather than given a fabricated limit. Supports -Location, -AllLocations and the
  standard subscription selectors. Reader-only. Resolves #7.
• Network quota visibility (Get-NetworkQuota.ps1) — a read-only collector that reports
  per-subscription, per-region Microsoft.Network usage vs limit (the data behind
  az network list-usages / the Microsoft.Network/locations/{loc}/usages API): Virtual Networks,
  Public IP Addresses, Network Interfaces, Load Balancers, NAT Gateways and the rest. Networking quota
  is a common, silent deployment blocker — you cannot create a VM if the subscription is out of public
  IPs or NICs in the region, regardless of compute headroom. Each counter is emitted with
  used / limit / available / PctUsed and NearLimit / AtLimit flags (threshold via
  -NearLimitPct, default 80). Counters the API returns with the placeholder 2147483647 limit are
  marked IsUnbounded (no fabricated math); per-VNet sub-counters (…PerVirtualNetwork) are flagged
  PerResourceScope. Supports -Location, -AllLocations, and the standard subscription selectors.
  Reader-only. Resolves #13.
• AKS scale-headroom check (Get-AksScaleHeadroom.ps1) — a read-only derivation that answers
  "if every node pool scaled to its autoscaler maxCount, would we run out of family quota?" It
  joins data the toolkit already reads — AKS node pools (Resource Graph), Microsoft.Compute/skus
  (VM size → family + vCPUs) and az vm list-usage (per-family used/limit) — to compute the
  incremental vCPUs each pool needs to reach its target, aggregates per VM family, and flags
  families that cannot fully scale within current quota. Pools in the same family are summed
  before comparison; Spot pools are checked against the separate regional low-priority pool, never
  the regular family quota; autoscale-disabled pools target their current count. Emits a per-pool
  detail CSV and a per-family rollup CSV, with Finding codes
  (OK / QuotaShortfall / MissingSkuMetadata / MissingQuotaFamily / SpotQuotaCheckNeeded).
  Reader-only. Resolves #12.
• Capacity Reservation inventory (Get-CapacityReservations.ps1) — a read-only collector that
  enumerates Microsoft.Compute/capacityReservationGroups and their child capacityReservations
  across visible subscriptions (ARM read, api-version 2024-07-01, $expand=instanceView) and reports,
  per reservation, the SKU, region, zone, reserved instance count, the capacity actually reserved at
  runtime, how many instances are consuming it, and idle / over-allocated / at-capacity flags. This
  closes the loop on the one construct that guarantees capacity (as distinct from quota), which
  docs/concepts.md already teaches. Reader-only; empty and unreadable (shared-scope) groups are
  surfaced as explicit rows. Resolves #11.
• Spot Placement Score lens (Get-SpotPlacementScore.ps1) — a read-only collector that calls the
  Azure Spot Placement Score API (Microsoft.Compute/locations/.../placementScores/spot/generate,
  api-version 2025-06-05) to return an allocation-likelihood signal (High / Medium / Low) per
  VM size × region × zone — the closest programmatic answer to the toolkit's "quota ≠ capacity"
  caveat. It scores Spot capacity (a proxy for regional pressure, not an on-demand guarantee), is
  time-sensitive (every row timestamped), and needs the read-only built-in "Compute Recommendations
  Role" in addition to Reader. Chunks requests to the API limits (≤8 regions × ≤5 sizes) and handles
  throttling / missing-role responses gracefully.
• Quota Group rollout (opt-in write tool) — Deploy-QuotaGroups.ps1, a generic,
  config-driven, idempotent engine that provisions Azure Quota Groups
  (Microsoft.Quota/groupQuotas) from a JSON design: registers providers, creates groups at
  a management-group scope, adds members, requests pooled group limits, and allocates quota to
  subscriptions. Requires PowerShell 7+, supports -WhatIf, and is guarded by ShouldProcess.
  This is the toolkit's only write capability and is kept separate from the read-only scans.
• Quota Group config bridge (New-QuotaGroupConfig.ps1) — turns a quota report into a
  deploy-ready design, auto-detecting either the toolkit-native wide quota-usage CSV (from
  Get-QuotaUsage.ps1) or an external azure-quota-reports CSV; computes allocations and
  pooled group limits with a configurable headroom buffer.
• Synthetic examples/quota-groups.sample.json design and a Quota Groups rollout
  guide; a "Services & coverage" matrix in the README.
• AI & agent governance section in AGENTS.md — states the toolkit runs no AI models and sends
  no data to any model or third party; the only AI relationship is an agent operating it under the
  file's read-only-by-default guardrails.
• Synthetic demo data generator (New-DemoDataset.ps1) — produces a complete,
  self-consistent fictional dataset ("Zava Inc") from a deterministic seed, with no Azure
  access required, so the dashboard can be previewed entirely offline. The generated universe
  exercises a range of dashboard states (SKU blocks and zone gaps, near-capacity quota, a GPU
  crunch, a pooled quota group, AKS lifecycle states, zone-redundant HA databases).

Changed

• Positioning reframed to "read-only by default." Docs, badges and templates no longer claim
  zero mutations now...

v0.1.0

First release of the Azure Capacity & Enablement Toolkit — read-only PowerShell scripts and a self-contained HTML dashboard for analyzing Azure regional/zonal capacity, SKU enablement, quota, and zonal resilience using only Reader access.

Highlights

• Capacity & enablement scans — discover which SKUs you run, check regional + zonal enablement, and see quota/headroom per family.
• Resilience view — per-subscription zone mapping, zone-pinned resource sweep, AKS and Flexible Server footprint.
• Region comparison — score candidate regions for a move or failover, with capacity-honest verdicts.
• One-command report — orchestrator joins the scans into a CSV + Markdown summary and an offline, tabbed HTML dashboard.

Everything is read-only — the only writes are local files under output/.

See the docs to get started, and the changelog for the full list.