v2.0.0 — Whitepaper-Aligned Decision Governance (Breaking)

Overview

RGDS v2.0.0 aligns the repository to the RGDS whitepaper’s decision-governance model.
This release strengthens RGDS as an auditable, non-agentic decision-support system by making governance requirements enforceable through schema + semantic validation.

This is a breaking release for v1.x decision logs.

---

Key Changes

Decision Log Requirements (now enforceable)

• Mandatory options analysis (≥2 options per decision)
• Explicit evidence completeness classification:
  • complete / partial / placeholder
• Residual risk captured as a first-class decision artifact
• Explicit risk posture declaration:
  • risk_minimizing / risk_neutral / risk_accepting
• Named human accountability (owner + approvals)
• Structured AI assistance disclosure when AI is used (tool identity, purpose, human review, overrides, risk assessment)

Schema, Template, and Validators

• Updated decision log schema (JSON + YAML) and template to prevent drift
• Strengthened semantic validation:
  • options minimum enforced
  • AI disclosure fields enforced when ai_assistance.used=true
• All canonical examples pass schema + semantic validation

---

Migration Notes (v1.x → v2.0.0)

Existing v1.x logs must be updated to conform to v2.0.0:

• add options_considered (≥2)
• add risk_posture enum
• add risk_assessment + residual risk statement/items
• add evidence item completeness_state
• ensure ai_assistance is present (and fully populated if used)

---

Governance Stance (unchanged)

• AI is explicitly non-agentic
• AI never decides, approves, or accepts risk
• Human accountability remains primary and explicit