v2.1.0 — Decision-Ready Governance
Global AI Governance Toolkit v2.1.0 turns an AI inventory record into policy-driven findings and a reproducible, human-reviewed AI Governance Decision Pack.
What Changed
- Critical governance findings now fail closed by default.
- The checked-in inventory schema is enforced before risk evaluation.
- The checked-in governance policy drives findings, severity, messages, and stable rule identifiers.
- Schema, policy, input, and configuration failures return exit code
2. - The one-command pipeline generates a deterministic Decision Pack when the selected gate permits continuation.
- Critical report-only runs preserve a visible blocked state.
- A checked-in reference Decision Pack is protected by regeneration and drift checks.
- Repository integrity validation covers Python, JSON, local Markdown links, action pins, manifest hashes, and generated-bytecode hygiene.
- Current documentation states the human-decision and evidence boundaries explicitly.
Decision Pack
The generated pack contains:
- Executive summary
- System profile
- Risk and findings
- Evidence and ownership
- Decision record
- Action plan
- Manifest
The generated decision status remains pending human decision.
Verification
The release gate runs the full runtime-alignment and Decision Pack test suite, repository validation, reference regeneration, documented quick start, valid path, blocked path, and explicit report-only path in GitHub Actions.
Evidence Boundary
This is a working public reference toolkit. It does not approve deployment, certify compliance, provide a legal or regulatory determination, accept risk, or replace legal, security, privacy, procurement, compliance, executive, or board review.