LUMINA-30 adoption pathway

Procurement and Vendor-Risk Prompt

A lightweight boundary prompt for AI products, AI agents, vendor-hosted workflows, and internal deployments.

This page is for procurement, vendor-risk, security, audit, and AI governance teams that need to ask a vendor or internal product owner a small number of boundary questions before approving an AI-mediated service, agent, workflow, or tool.

It does not replace legal, regulatory, security, privacy, or safety review. It adds a narrow LUMINA-30 boundary prompt: whether effective human refusal, suspension, review, and rollback remain possible before an AI-mediated process becomes irreversible.

Minimum vendor prompt

Before this AI-mediated process changes credentials, executes tools, modifies production state, removes recovery options, or creates irreversible commitments, who can refuse, stop, suspend, review, or roll it back, and what evidence will show that this authority existed before irreversibility?

Four fields to request

FieldAsk for
ActorWhich human, vendor team, internal owner, AI assistant, automation layer, or privileged system API can initiate or execute the action?
AuthorityWho has the authority to refuse, suspend, approve, or roll back the action before irreversibility?
EvidenceWhat logs, approvals, alerts, tickets, or audit records show that the authority existed and was usable in time?
Irreversibility windowWhen does the process become difficult or impossible to reverse, recover, or meaningfully review?

When to use this prompt

Safe use

A vendor or internal team may answer this prompt without claiming LUMINA-30 certification, approval, compliance, or safety guarantee. The safer description is:

This review refers to LUMINA-30 to check whether effective human refusal and rollback authority remain available before irreversible escalation.

Next routes

One-Question Pilot

Start with one low-risk review question.

Adoption Review

Return to the adoption-review entry point.

Public-source review template

Separate public facts, inference, missing evidence, and unresolved conclusions.