Skip to content

OpenClaw Enterprise Admin Guide

This guide covers the administration of OpenClaw Enterprise (codename: Redclaw), the enterprise extension layer for the OpenClaw open-source AI assistant. OpenClaw Enterprise is self-hosted and open-source, designed around a plugin-first architecture that extends OpenClaw via plugins rather than forking.

Core Admin Capabilities

OpenClaw Enterprise provides the following administrative capabilities:

Capability Description Guide
Policy Engine Hierarchical OPA-based policy engine with 7 domains and graduated autonomy Policy Engine
Data Classification Four-level classification system with propagation and model routing Data Classification
Connectors OAuth-based integrations with Gmail, GCal, Jira, GitHub, GDrive Connectors
Audit Logging Immutable, append-only audit trail for all actions Audit Log
RBAC and SSO OIDC-based SSO with four built-in roles RBAC and SSO

Architecture Overview

OpenClaw Enterprise follows several key architectural principles:

  • Upstream First -- All functionality is delivered via the OpenClaw plugin API (registerTool, registerHook, registerService, registerHttpRoute, registerGatewayMethod, registerContextEngine). The upstream OpenClaw codebase is never forked.
  • Policy Over Code -- Behavior is defined by a hierarchical policy engine (OPA with Rego policies), not hardcoded logic. Administrators control what the system can and cannot do through policy, not source code changes.
  • Least Privilege By Default -- All actions are deny-by-default. Connectors start read-only. The policy engine fails closed (if OPA is unreachable, all actions are denied).
  • Data Classification Travels With Data -- Every piece of data carries its classification level. Summaries, derivatives, and agent-to-agent exchanges all inherit classification from their source material.

Deployment Components

A production OpenClaw Enterprise deployment includes:

  • 15 TypeScript plugins running on Node.js >= 22 (strict mode)
  • 1 Go-based Kubernetes operator managing CRDs, reconcilers, webhooks, and RBAC
  • OPA sidecar for policy evaluation (localhost:8181)
  • PostgreSQL database (no SQLite in production) with 7 migrations
  • 7 Rego policy files covering all policy domains

Getting Started

  1. Deploy the Kubernetes operator and CRDs (see operator documentation)
  2. Configure SSO/OIDC authentication with your identity provider
  3. Set up enterprise policies for your organization
  4. Configure data classification defaults
  5. Enable and configure connectors for your integrations
  6. Verify audit logging is operational

Quick Reference: Admin API Endpoints

Endpoint Method Description Required Role
/api/v1/policies GET, POST, PUT, DELETE Policy management org_admin+
/api/v1/connectors GET Connector status and configuration org_admin+
/api/v1/audit GET Query audit log org_admin+
/api/v1/audit/export GET Export audit data enterprise_admin
/api/v1/admin/tenants GET, POST, PUT Tenant management enterprise_admin
/api/v1/admin/status GET System status enterprise_admin
/api/v1/admin/metrics GET System metrics enterprise_admin