RGDS v2.0.0 — Whitepaper-Aligned Decision Governance (Docs, Schema, Examples, Validators)

[mj3b]

Summary

This PR updates RGDS to v2.0.0, fully aligning the repository with the RGDS whitepaper’s decision-governance model.

The changes are intentionally comprehensive: documentation, schema, examples, and validation logic are now synchronized so that governance intent is enforced, not just described.

This is a non-agentic, human-governed update focused on decision defensibility, auditability, and regulatory credibility.

---

What Changed

1. Documentation (Governance First)

• Clarified RGDS as a decision governance system, not a narrative memo or agent framework
• Added explicit requirements for:
  • Mandatory options analysis
  • Evidence completeness classification (complete / partial / placeholder)
  • Residual risk documentation
  • Named human accountability
  • AI assistance disclosure when applicable

2. Decision Log Schema & Template

• Introduced required fields for:
  • options_considered (≥2 options enforced semantically)
  • evidence_completeness
  • risk_posture
  • risk_assessment
  • ai_assistance (always present; fully structured when used)
• Updated YAML/JSON schema and templates to prevent drift

3. Canonical Example Decisions

• All example decision logs updated to conform to v2.0.0 schema
• Added explicit evidence completeness at the item level
• Expanded AI-assisted examples to demonstrate full disclosure and human override patterns

4. Validation & Enforcement

• Strengthened semantic validation:
  • Enforces ≥2 options
  • Enforces AI disclosure requirements when AI is used
  • Preserves warnings vs errors to avoid reviewer fatigue
• All examples pass schema + semantic validation

---

Why This Matters

This update closes the gap between:

• “Good documentation” and
• Enforceable, auditable decision governance

It ensures that:

• Decisions cannot be recorded without explicit tradeoffs
• Risk acceptance is visible and owned
• AI involvement is disclosed, bounded, and reviewable
• The repo can be trusted as a reference implementation in regulated contexts

---

Review Guidance

Recommended review order:

1. README.md and docs/decision-log.md (governance intent)
2. decision-log/decision-log.schema.* (enforced requirements)
3. examples/ (canonical decision patterns)
4. scripts/validate_decision_log.py (semantic rules)

This PR is large by design but logically cohesive.

---

Backward Compatibility

• This is a breaking change for decision logs created under v1.x
• Migration is intentional and documented; v1.x examples remain historically informative but not schema-valid

---

Status

• Documentation aligned
• Schema and templates updated
• Examples validated
• Validators passing

---

Checklist

• Docs updated
• Schema updated
• Examples updated
• Validation passing
• No autonomous or agentic behavior introduced