Standards conflict scenarios (worked examples)

Illustrative §5 Standards traceability outcomes. Control IDs are fictional; map to your registry.example.yaml shape.


1. Security — customer control vs fast iteration (L1 beats L6)

Situation: A Spark proposes copying a production database snapshot to a developer laptop for debugging. Discipline heuristic (L6) says “developers need realistic data.” Customer DPA (L1) forbids production PII outside production.

Field Example content
standards_considered L1 CUST-ACME-DPA-7; L5 secure engineering practice (realistic test data); L6 informal team habit
standards_applied L1 wins — no production PII in non-prod
conflicts_detected L6 habit vs L1 contractual control
waivers_required none (waivers cannot waive statute/DPA without legal path; use masking/subset instead)
evidence_obligations Synthetic data pipeline or masked export; ticket link showing data-class compliance; optional Compliance Versona pass on Ingot

Top-level recommendation: Rework or Proceed with conditions (conditions: approved synthetic data only).


2. Compliance — regulation vs repo convenience (L1 beats L3)

Situation: Team wants one shared S3 bucket for all environments to cut cost ( L3 local convenience). Regulatory mapping (L1) requires logical separation and access logging per environment for audit.

Field Example content
standards_considered L1 sector retention/access rules; L3 proposed bucket consolidation ADR draft
standards_applied L1 — separate buckets or equivalent controls with demonstrable separation
conflicts_detected L3 cost/convenience vs L1 separation of environments
waivers_required If org legal accepts compensating controls: file waiver with compensating_controls; else none and change design
evidence_obligations Architecture diagram; logging/retention config references; Assay Gate checklist

Handoff: Architecture + DevOps Versonas to implement L1-compliant design.


3. Product — roadmap pressure vs Assay evidence (L4 beats L6)

Situation: Product wants to ship a Spark on Friday (L6 urgency). Forge Assay pattern (L4) says verify Sparks lack open significant concerns without mitigation. No L1/L2 conflict.

Field Example content
standards_considered L4 Assay Gate / evidence; L5 PM discipline (market timing)
standards_applied L4 — do not bypass evidence for “just this once” without Ember Log risk acceptance
conflicts_detected L6 schedule pressure vs L4 methodology
waivers_required If shipping with accepted risk: Ember Log + optional waivers_required pointing to governance approval; else none and defer
evidence_obligations Minimal agreed test/evidence for the Spark type; PM articulates trade-off in Ember Log

Top-level recommendation: Proceed with conditions (explicit evidence + logged trade-off) or Bank until evidence exists.


4. Engineering — OWASP-style practice vs org crypto standard (L2 beats L5)

Situation: Software Engineering Versona (L5) suggests a common JWT-in-localStorage pattern for speed. Org crypto baseline (L2) mandates httpOnly cookies for session tokens on customer-facing apps.

Field Example content
standards_considered L2 ORG-SEC-014-class transport/session rules; L5 common SPA patterns
standards_applied L2 — session storage mechanism must meet org standard
conflicts_detected L5 convenience pattern vs L2 mandated control
waivers_required none unless Architecture files exception via GRC
evidence_obligations Code review note; Security Versona spot-check; link to org standard in PR

Top-level recommendation: Rework implementation approach or Proceed with conditions (if prototype is throwaway and not customer-facing — state boundary clearly).


Summary

  • L1–L2 almost always defeat L4–L6 when they collide on the same requirement.
  • L4 Forge beats L6 ad-hoc urgency for evidence and ceremony expectations.
  • Waiver records document approved deviation from a specific control_id; they do not “cancel” law.

See Standards precedence — which controls win for the full stack.