The fail-closed governance layer for AI execution.
Gvner is the decision plane between agent intent and production execution. It enforces policy, records every decision, and now supports a unified integration contract across runtime families.
Setup Docs Map
Use this sequence to implement and operate a full tenant rollout.
1. Tenant Quickstart
Create baseline tenant setup and run first governed flow.
2. Choose Integration Path
Pick unified adapter (recommended) or compatibility adapters.
2b. Read Adapter Docs
Review per-adapter purpose, inputs, policy touchpoints, and console monitoring paths.
3. Wire Runtime Adapter
Implement /adapters/http/* register, intent, heartbeat, and runtime inventory.
4. Validate API Contract
Use current endpoint contracts and fail-closed behavior.
5. Configure Identity
Set up SSO, SCIM, roles, and session policy with the Identity guides.
6. Define Policy + Approvals
Configure policy rules and approval workflows before enforcement.
7. Enable Evidence + Exports
Generate immutable evidence and export packets/receipts.
8. Operate in Console
Run governance, security, integrations, and ops checks daily.
Unified Integrations
New and non-native runtimes use one canonical adapter contract. Existing OpenClaw and Temporal paths remain stable.
Canonical Adapter API
Use /adapters/http/register, /adapters/http/intent, /adapters/http/runtimes, and /adapters/http/heartbeat.
Integration Families
LangGraph, Semantic Kernel, Bedrock, Vertex ADK, automation platforms, and worker queues all map to the same governance core.
Top 20 Adapter Directory
GitHub, AWS, Kubernetes, Stripe, OpenAI, Anthropic, trading, blockchain, and more under one adapter contract.
OpenClaw Compatibility
OpenClaw keeps dedicated compatibility endpoints for existing deployments.
Temporal Compatibility
Temporal keeps activity-level compatibility endpoint behavior unchanged.
Core Principle
This boundary is enforced in code, visible in the console, and backed by immutable evidence.
What Gvner Does
External Execution Authority
Gvner sits between AI agent intent and production systems.
Policy & Approval Governance
Execution is checked against policy, budget, and approval requirements.
Evidence & Audit Exports
Every decision creates audit-grade evidence and exportable artifacts.
Identity Governance
SSO, SCIM, sessions, and role mapping are governed and auditable.
Agent Governance
Track agent behavior, baselines, anomalies, and enforcement.
Operational Integrity
Operate with live readiness, resilience checks, and SLA oversight.
Who Uses Gvner
Gvner is for teams running agents where mistakes have real business impact:
Typical users include:
Each group sees the same source of truth: decisions, evidence, and operational state.
What Gvner Is Not
Gvner does not replace intelligence. Gvner governs execution. This boundary is intentional.
How Gvner Fits
An agent emits an intent
The runtime routes intent to Gvner adapter endpoints
Gvner evaluates policy, identity, budget, and risk
Gvner returns ALLOW or DENY with a recorded decision
Only ALLOW decisions can proceed to execution
Agents do not bypass this gate.
Gvner integrates through a unified adapter core plus stable compatibility surfaces for OpenClaw and Temporal.
Provable by Design
Every decision is packaged as verifiable evidence:
Auditors can validate outputs without internal system access.
Evidence generation is default behavior, not an optional add-on.
Built for Oversight
Oversight is product behavior: visible, testable, and repeatable.
Getting Started
Adopt Gvner in a staged rollout:
Step 1: How It Works
Understand the decision boundary and execution flow.
Step 2: Setup Docs
Use Quickstart, Identity, and Policy guides.
Step 3: Integration Hub
Pick unified adapter (recommended) or OpenClaw/Temporal compatibility paths.
Step 4: Console Operations
Operate via console runbooks and monitor integration runtime status.
Move from visibility to enforcement at your own pace.
Call to Action
Ship agent systems with explicit control, measurable reliability, and audit-ready evidence.