workbook-generator

name

workbook-generator

description

Discovery-linked Azure Monitor workbook generation. Use when you have an SRE Agent resource discovery result and need to turn it into deployable Azure Monitor (Log Analytics) workbooks for the infrastructure, application, network, and storage domains. Triggers: "auto-generate workbooks", "discovery-linked dashboards", "build workbooks from inventory", "operational dashboards for a workspace". This skill consumes a discovered inventory (it does not perform discovery — that is the SRE Agent's built-in capability) and emits workbook JSON ready for the Bicep or Terraform deployment paths. Invoked by the sre-workbook-orchestrator agent, but usable standalone.

Workbook Generator

Turns an SRE Agent discovery result into Azure Monitor workbooks aligned to what is actually deployed. Discovery itself is the SRE Agent's built-in capability; this skill consumes its output. Implements the generation half of Story #33 — Auto-Generated Workbooks (Task #18, Adapt Workbook Generation Logic).

When to use

• A customer needs operational dashboards stood up quickly from their real estate.
• The discovered inventory has changed and dashboards must be re-aligned to it.
• You need workbook JSON to feed the bicep/ or terraform/ deployment.

Inputs

Input	Description
SRE Agent discovery result (JSON)	Output of the agent's built-in discovery. Either a flat resources[] list or a pre-bucketed domains{} inventory. See generator/sample-discovery-input.json.
Target Log Analytics workspace resource ID	Where the workbooks will resolve queries.
Domains (optional)	Subset of infrastructure,application,network,storage. Default: all with discovered resources.

Procedure

Stage 1 — Consume the discovery result

Provide the SRE Agent's discovery output as JSON. The generator buckets a flat resources[] list into the four domains by resource type (case-insensitive), or uses an already domain-bucketed inventory as-is:

• infrastructure — VMs, VMSS, Arc machines, availability/scale sets
• application — App Service, Functions, Container Apps, AKS
• network — VNets, NSGs, Load Balancers, App/Front Door, Public IPs
• storage — Storage accounts, Managed Disks, SQL, Cosmos

Stage 2 — Generate

Bind the inventory to the domain templates and emit one workbook per populated domain:

python ./generator/generate_workbooks.py \
  --inventory ./discovery-result.json \
  --workspace-resource-id <workspace-id> \
  --templates-dir ./templates \
  --output-dir ./out

Generation rules:

• Skip empty domains. No storage accounts discovered -> no storage workbook.
• Bind the workspace. Every template's {{workspaceResourceId}} and {{crossComponentResources}} placeholders are replaced from the inventory.
• Tag provenance. Each workbook header records the inventory timestamp and source subscription so dashboards are auditable.
• Deterministic output. Re-running on the same inventory yields byte-identical workbooks (stable ordering), so deployments are idempotent.

Stage 3 — Hand off

Emit out/<domain>.workbook.json. These are the deployment inputs for:

• Bicep: bicep/main.bicep (parameter workbooks[], loaded in main.bicepparam)
• Terraform: terraform variable workbooks_dir (reads *.workbook.json)

Output contract

• Files conform to Azure Monitor Workbook schema Notebook/1.0.
• No unbound {{...}} placeholders remain.
• A manifest.json lists generated domains, source inventory hash, and target workspace for the orchestrator's Validate stage.

Definition of done

• Generated workbooks for every domain with discovered resources.
• All queries scoped to the discovered workspace (no manual edits — AC #3).
• Output consumable unchanged by both Bicep and Terraform paths.