Help center

Use this when you need a fast internal readout before a Microsoft AI agent pilot, customer review, or trainer-led workshop.

1. Answer with the current truth

Choose None, Ad hoc, Partial, or Ready for each control. Do not answer for the target architecture unless it is already implemented and reviewed.

2. Read the score as readiness, not certification

The score is a practical launch-readiness signal. It maps common agent risks to DORA, NIST CSF, and CIS language so security and platform teams can discuss evidence without turning the scanner into a legal opinion.

3. Copy the report or request a review

Copy report creates a plain-text summary for internal notes.
Request review opens an email draft with the current report.
The work-email form saves the request server-side and then opens the same email draft.

4. Good first fixes

For most teams the first fixes are identity ownership, least-privilege tool permissions, audit logging, incident handling, and a concise evidence pack for the pilot owner.

Run the scanner