Aller au contenu principal

Human Interaction

Use this page when you need to understand where human approval, escalation, or override is still mandatory in Helpifyr workflows.

When to use this page​

  • You are designing a workflow with human review.
  • You need to distinguish ordinary approval from break-glass intervention.
  • You need public-safe guidance for escalation behavior.
  • You need to understand where implementation, approval, and final closeout separate.

Prerequisites​

Architecture / Flow​

Step-by-step procedure​

1. Distinguish the interaction type​

  • approval:
    • normal review gate for a bounded step
  • escalation:
    • raise to a different owner or authority
  • override:
    • controlled exception to a normal rule
  • break-glass:
    • exceptional path for urgent recovery, still evidence-bound

2. Base the decision on readback​

Illustrative reads:

GET /api/v1/signoff/readiness
GET /api/v1/security/readiness
GET /api/v1/recovery/readiness

3. Keep authority and scope explicit​

Human review should answer:

  • who may approve this action
  • what exact state was reviewed
  • what action is allowed next

4. Re-verify after the human step​

Even when approval was valid, the workflow is not complete until the resulting state is read back.

Verification​

This page is being applied correctly when:

  1. the interaction type is explicit
  2. approval is tied to current evidence
  3. break-glass remains exceptional rather than normal

Common failure modes​

Using human judgment to replace missing evidence​

Problem:

  • the workflow becomes arbitrary.

Better path:

  • escalate or block when evidence is insufficient

Treating break-glass as a normal operator shortcut​

Problem:

  • safety posture erodes over time.

Better path:

  • keep break-glass separate, auditable, and rare

Source Truth​

Next paths​