OECD Crosswalk Use Cases Tool Card API Records About Main Site Stable Tool

Open-source pre-execution governance gateway

Before an AI-enabled action becomes a governed effect, ask whether it may execute.

May this proposed action be authorized to produce a governed effect now?

Policy profilesRisk levelEvidence pack hashHuman reviewAPI gate ready

MayExecute v2.0
Governance Gateway Before Governed Effect

Verified Mandate

Valid Constraints

Live-Context Integrity

Accountability

Reviewability

Sufficient Verifiable Proof

↓

PROCEED / REVIEW / HOLD / BLOCK

+ LOW / MEDIUM / HIGH / CRITICAL

Execution Governance Reference Specification v0.9.7.3

Pre-Effect Authorization Interface and Effect Authorization Receipt

This dev preview is an illustrative implementation preview for EG-style Proceed / Review / Hold / Block checks. It is not certification, not a conformance claim, and not authority to execute.

Open on Zenodo · Reference Page · Schema Test

Evaluate a proposed action

Quick-start template

Templates pre-fill the form using common AI, IoT, robotics, property, procurement, public-service, payment, and governance scenarios. Review and edit all fields before generating a record.

Policy profile

v2.0 applies different risk weights and critical failure rules by profile.

Action title

System or context

Proposed action

Expected governed effect

Evidence reference / hash / URL / ticket

Evidence files for evidence pack hash

Files are hashed in memory and not stored.

Privacy mode

Standard mode: store submitted text Hash-only privacy mode

Six-condition assessment

1. Verified Mandate

Is there a verified mandate? Who or what gives authority for the action?

Yes Unclear No

2. Valid Constraints

Are the rules and limits valid now? Do current policies, limits, laws, safety rules, or operational constraints permit the action?

Yes Unclear No

3. Live-Context Integrity

Is the live context still intact? Has the situation drifted since the instruction, recommendation, or AI output was generated?

Yes Unclear No

4. Accountability

Who remains accountable? Is there a responsible person, team, system owner, or authority chain?

Yes Unclear No

5. Reviewability

Can the basis be reviewed? Can a reviewer understand why the action was classified as proceed, review, hold, or block?

Yes Unclear No

6. Sufficient Verifiable Proof

Is there sufficient verifiable proof? Are there evidence references, hashes, files, logs, or attestations that can be checked?

Yes Unclear No

Human review / sign-off

Reviewer name

Reviewer role

Review note

Development build — v2.0.0