M25 · Deployment, testing, and agent guardrails
Test 1 - Mike's Review · 13 questions · printed Oct 21, 2026
25.0 Deployment approach: full or incremental, sandbox chain, who deploys to production, change windows, and release freezes.
Pending
25.1 Build authority: complete the table in Part D so the agent knows what it may do alone, what needs approval, and what it may never do.
Pending
25.2 Activation policy: Revenue Cloud components have draft, active, and inactive states. May the agent activate pricing procedures, constraint models, approval workflows, and object state definitions in sandboxes? In production?
Pending
25.3 Test data: which records may the agent create, what naming prefix marks them, and is production data excluded or masked?
Pending
25.4 Acceptance scenarios: at least five per unlocked module, in the Part D format.
Files: C-33 (not uploaded)
Pending
25.5 Regression: existing processes and integrations that must keep working.
Pending
25.6 Rollback: how to revert (deactivate, restore a version, restore data), and what backup must exist before each data load.
Pending
25.7 Logs the agent must produce: build log, assumptions list, and deviations from this questionnaire. State format and location.
Pending
25.8 Stop conditions: when must the agent halt and ask (an unsupported feature, a limit exceeded, a contradictory rule, a failed acceptance test)?
Pending
25.9 Deliverables: design document, data dictionary, admin runbook, and their formats.
Pending
25.10 After go-live: who owns changes, how long is hypercare, and what is the defect response time?
Pending
25.11 Salesforce release timing: any freeze around the next release, and will you test in a preview sandbox first?
Pending
25.12 Who signs off, and on what evidence?
Pending