# GovernWorld — Enterprise AI Governance & Safety Gateway

> GovernWorld is the Enterprise AI Governance Enforcement Gateway. It provides bidirectional detection, policy evaluation, stateful transformations (PII/PHI redaction, prompt injection guardrails, toxic/harmful content filtering), and cryptographic audit evidence for AI systems, LLM agents, Model Context Protocol (MCP) tools, and enterprise workflows.

## System Overview & Architecture

GovernWorld operates as an in-process and edge-deployed governance gateway with fail-closed runtime semantics (`DENY | REQUIRE_REVIEW | DEGRADE | RETRY | ABORT`).

- Platform Architecture: 13-stage execution pipeline (received -> authenticated -> tenant-resolved -> normalized -> classified -> policy-evaluated -> transformed -> egress-authorized -> dispatched -> output-received -> output-scanned -> output-transformed -> evidence-committed -> completed).
- Security Boundary: Step-up authentication (5-minute hard TTL), dual-approval workflows, break-glass audit logs, and hardware/KMS-backed audit evidence stores.
- Storage: Managed PostgreSQL (Flexible Server) for state and audit outbox, Upstash Redis for rate-limiting, BYOS for encrypted evidence persistence.

## Key Features & Capabilities

- **Bidirectional Detection & Redaction**: Detects and redacts 50+ PII/PHI data types (SSN, credit card, medical record numbers, IBAN, API keys, JWTs) across text, JSON, and document formats before outbound dispatch to third-party LLMs.
- **Model Context Protocol (MCP) Tool Allowlisting**: Strict runtime authorization gates for MCP tools, restricting LLM agent access to approved schemas, function calls, and data sinks.
- **Break-Glass Emergency Protocol**: Immutable audit ledger and approval enforcement for elevated privileges and administrative overrides.
- **Enterprise SSO & Identity Governance**: OIDC, SAML 2.0, Microsoft Entra ID (Azure AD), and OAuth 2.0 integration with role-aware console RBAC.
- **Evidence & Audit Assurance**: Merkle-tree chained cryptographic evidence records committing all prompt inputs, tool calls, policy decisions, and output transformations.

## Core Navigation & Documentation Links

- [Platform Overview](https://governworld.acrlogic.com/platform): Complete overview of the runtime governance gateway architecture and deployment models.
- [Runtime Governance Pipeline](https://governworld.acrlogic.com/runtime-governance): Detailed documentation of the 13-stage fail-closed execution engine.
- [Security & Compliance Specification](https://governworld.acrlogic.com/security): Security posture, encryption standards, threat modeling, and SOC2/HIPAA/ISO27001 compliance framework.
- [MCP & AI Agent Governance](https://governworld.acrlogic.com/ai-agents): Tool allowlisting, function call scanning, and egress filtering for autonomous agents.
- [Solutions & Enterprise Deployment](https://governworld.acrlogic.com/solutions/enterprise): Infrastructure blueprints for finance, healthcare, government, and legal enterprise environments.
- [SDK & Integration Documentation](https://governworld.acrlogic.com/sdk): Developer guides, API keys, TypeScript/Python SDK usage, and code samples.
- [Full LLM Specification](https://governworld.acrlogic.com/llms-full.txt): Complete un-truncated machine-readable architecture and API manifest for deep context ingestion.