Jev in Cloud Operations with PowerShell (Olivier Miossec)
External DEV write-up by Olivier Miossec (2026-10-01): use Doug Finke's Jev PowerShell module to gate Bicep What-If changes, review NSG rules, and triage Azure Policy linter findings with hosted TypeSafe Choice / Noul / Score questions.
What this guide points to
Olivier Miossec published Decisions/System One model in action, Jev in Cloud operation with PowerShell on DEV (posted 2026-10-01, edited 2026-10-08). The article walks through everyday Azure cloud-engineer tasks where answers are enumerable and should trigger a scripted next step. It uses the community dfinke/Jev PowerShell module (New-JevQuestion, New-JevYesNoQuestion, Invoke-Jev) against hosted TypeSafe Jev with TYPESAFE_API_KEY.
Worked examples in the post and companion repo:
- Bicep What-If gate — strip
NoChange/Ignore, summarize deltas, then ask Jev whether the plan weakens security, exposes a public IP, and what risk score to assign; code maps thresholds into Pass vs Review. - NSG review — PowerShell owns deterministic findings (management ports open to the internet); Jev scores intent, compliance to a written standard, breadth, temporary look, description mismatch, and severity; actions such as
security-ticket-p1stay in code. - Azure Policy linter triage — run Azure/azure-policy-linter, then ask Jev whether findings matter, whether descriptions match the rule, and whether assignment could weaken security.
The author reports the article’s examples used 103,525 tokens for about $0.0039 total. Sample code lives in omiossec/Jev-PowerShell-Cloud-Eng (AGPL-3.0).
This page is an unofficial handbook pointer — verify claims against the article, the module README, and TypeSafe docs. Thresholds are demo-tuned.
Takeaways
- Keep hard security rules in PowerShell; ask Jev only for judgment that is fuzzy but still closed-set.
- Put allow / review / ticket actions in code after Jev returns probabilities.
- Shrink IaC plan or policy JSON before the call so state stays within context.