Plan Manifest Schema

Reopenable projects

Save project downloads a separate project format with kind: "azurelocal-surveyor-project", numeric schemaVersion: 1, stateVersion: 11, appVersion, name, savedAt, planningArea and inputs. The optional area tag is storage or workload; legacy files without it default to workload planning. State version 11 (2.9.0) adds the optional hardware.rackAware flag and the four-way-mirror resiliency value; both are additive, so version 10 files open unchanged and schemaVersion stays 1. Importers that read schemaVersion keep working; they should map four-way-mirror and note rackAware.

Open projects in their matching area using Open / Compare. Preview comparisons before restoring. Storage projects exclude workload demand. The existing manifest format below remains supported in Workload Planning. See the planning guide for explicit transfers between areas.

Report manifests

The Export JSON button on the Reports page produces a SurveyorPlan JSON file. This file is the primary handoff artifact for Ranger and any downstream automation.

File name

azure-local-plan-YYYY-MM-DD.json

Schema versioning

The schemaVersion field uses "MAJOR.MINOR" string format.

• Ranger (and any other consumer) checks only the major version: schemaVersion.split('.')[0] === '1'
• MAJOR bumps when a field is removed or its type changes incompatibly — consumers must reject the file
• MINOR bumps when new optional fields are added — consumers may safely ignore unknown fields

The current version is "1.0".

Top-level structure

{
  "schemaVersion": "1.0",
  "surveyorVersion": "1.4.0",
  "generatedAt": "2026-04-14T12:00:00.000Z",
  "provenance": { ... },
  "inputs": { ... },
  "outputs": { ... }
}

Field	Type	Description
schemaVersion	"1.0"	Manifest schema version — check major only
surveyorVersion	string	Surveyor app version that generated the file
generatedAt	string	ISO 8601 UTC timestamp of export
provenance	object	How and by whom the plan was created
inputs	object	Verbatim snapshot of all user-entered inputs
outputs	object	Frozen computed results at export time

provenance

{
  "source": "manual",
  "notes": "Pre-deployment sizing for customer Site A",
  "approvedBy": "engineer@example.com",
  "approvedAt": "2026-04-14T15:30:00.000Z",
  "baselineLabel": "Pre-deployment baseline — Site A"
}

Field	Type	Required	Description
source	"manual" | "imported" | "url-share"	yes	How inputs were created
notes	string	no	Free-text annotation
approvedBy	string	no	Approver name or email — set by Ranger or hand-edit
approvedAt	string	no	ISO 8601 UTC approval timestamp
baselineLabel	string	no	Human-readable baseline label for Ranger display

For v1.7.0 there is no UI for stamping approval. Set approvedBy, approvedAt, and baselineLabel by editing the JSON directly or via Ranger.

inputs

A verbatim copy of all Zustand store slices at export time. All type definitions live in src/engine/types.ts and use kebab-case enum values (e.g. "three-way-mirror", not "Three-Way Mirror").

Field	Type	Description
hardware	HardwareInputs	Cluster node, drive, CPU, and RAM config
advanced	AdvancedSettings	Efficiency factor, infra volume, overrides
volumes	VolumeSpec[]	User-defined planned volumes
volumeMode	"workload" | "generic"	Volume suggestion mode
avdEnabled	boolean	Whether AVD scenario is active
avd	AvdInputs	AVD session host and FSLogix inputs
sofsEnabled	boolean	Whether SOFS scenario is active
sofs	SofsInputs	SOFS guest cluster inputs
mabsEnabled	boolean	Whether MABS scenario is active
mabs	MabsInputs	MABS backup server inputs
aks	AksInputs	AKS cluster inputs (aks.enabled is the toggle)
virtualMachines	VmScenario	VM workload inputs (virtualMachines.enabled is the toggle)
servicePresets	ServicePresetInstance[]	Arc-enabled service preset instances
customWorkloads	CustomWorkload[]	User-defined custom workload entries

outputs

Computed at export time and frozen. Ranger can consume these directly without re-running the engine. Per-scenario fields (avd, sofs, aks, mabs) are only present when the scenario was enabled at export time.

Field	Type	Always present	Description
capacity	CapacityResult	yes	Raw pool, usable, reserve, effective usable
compute	ComputeResult	yes	vCPU, memory, N+1 failover analysis
volumeSummary	VolumeSummaryResult	yes	Planned volumes, utilization
workloadTotals	WorkloadSummaryResult	yes	Aggregate vCPU, memory, storage across all scenarios
health	HealthCheckResult	yes	Volume and compute health check results
avd	AvdResult	when avdEnabled	Session host count, storage breakdown
sofs	SofsResult	when sofsEnabled	Profile demand, IOPS, auto-size result
aks	AksResult	when aks.enabled	Control plane + worker node totals
mabs	MabsResult	when mabsEnabled	Backup volume sizing, internal mirror factor

Ranger integration notes

1. Version check first: reject files where schemaVersion.split('.')[0] !== '1'
2. Trust outputs: the outputs section is a frozen snapshot — do not re-run the engine; use these values directly
3. Baseline approval: set provenance.approvedBy, provenance.approvedAt, and provenance.baselineLabel before importing into Ranger as an approved baseline
4. Enum values: all ResiliencyType, DriveMedia, and similar enum fields use kebab-case — "three-way-mirror" not "Three-Way Mirror"

Round-trip import

A plan exported as JSON can be loaded back into Surveyor in a future release. When loaded, provenance.source will be set to "imported" to distinguish it from a freshly-typed plan. This feature is not yet implemented in v1.7.0.

JSON Schemas

Machine-readable, versioned schemas ship in this folder and are checked by tests (src/state/schema.test.ts):

• project.schema.json — the saved project (schemaVersion 1, stateVersion 10 or 11).
• plan-manifest.schema.json — the Export JSON manifest (schemaVersion 1.x; 2.9.0 adds the optional targetRelease).

Both use JSON Schema 2020-12. The hardware block is strict (unknown keys are rejected) so typos are caught; other blocks allow additional properties. A schema change that rejects a previously valid file requires a new schemaVersion major.