Walks an avd-current-state assessment step by step. Run each command in Azure Cloud Shell (PowerShell), paste the output back here, and answer the discovery questions. Everything happens in your browser: nothing you paste or type is sent anywhere, and every command only reads.
Run the commands in
Assessment settings
Used to build the commands. Kept only in this browser.
Clears the saved progress, answers, notes and settings from this browser.
Collect the inventory stage 1 of 5
The collector reads what is in your subscriptions through Resource Graph and writes an evidence package in Cloud Shell. Then the facts script gives the estate a pseudonym, which the later stages use instead of any name. Both only read.
Run in Cloud Shell
Collect and pseudonymize
Running it on a schedule? Check the optional reader
An optional read-only identity lets GitHub Actions in your private repo run the collector with nobody signed in. Set it up from the reader guide; this check only reads.
Check the reader
Paste the output
Read in this browser. The paste itself isn't saved.
Discovery interview stage 2 of 5
What telemetry can't know. The answers become intake.json (avd-intake/v1), which holds only choices, region names, times and the estate pseudonym. The narrative goes in your private notes, a file that stays on your device.
The validator checks every host pool after deployment: wiring, app group access, session host health, profile storage, scaling, diagnostics and backup. Scoring against the avd-assess method, with intake.json, joins this step when the validator's assess mode lands.
Run in Cloud Shell
Validate
Paste the output
Read in this browser. The paste itself isn't saved.
Review the findings stage 4 of 5
Decide each finding: fix it, accept it with a reason, or defer it. The decisions become review.json (avd-assessment-review/v1), which pins the report by its hash and holds only check IDs and reason codes. A note on a finding goes to your private notes file.
The findings you chose to fix, and every finding nobody reviewed, become an ordered plan: what good looks like and why, the landing zone parameter or setting that changes it where one is known, and what is outside the landing zone's reach. Accepted and deferred findings stay out. Saved as plan.json (avd-remediation-plan/v1), which holds only check IDs, citations and public values, and plan.md for people.
Load assessment.json in the Review stage first.
The plan, in priority order: gap, warn, unknown; then severity
Order
Check
Outcome
Change
Why
A landing zone change fixes only what the landing zone deployed. A finding on a resource it doesn't own is changed where that resource is managed.
Run in Cloud Shell
Writes plan.json beside the evidence packages. It holds no names, so this block is safe to paste.
