Home

GRDL — Governance Rule Definition Language

GRDL is a deterministic governance policy engine for autonomous AI agents. Write rules in YAML, compile to any enforcement backend, evaluate agent decisions in sub-microsecond latency.

Runtime-agnostic. One ruleset governs agents in OpenShell, Docker, Kubernetes, or standalone HTTP.

Why GRDL exists

Autonomous AI agents — Gemma 4, OpenClaw, Claude Code, LangChain agents — make decisions and take actions without human input. The industry has infrastructure security (OpenShell sandboxes, Docker isolation). Nobody has decision governance: should this agent spend beyond its budget? Can it escalate its own permissions? Is the delegation chain deep enough to cause cascading failures?

GRDL fills this gap. It is the YAML standard for AI agent governance.

Wiki contents

Page	Description
Getting Started	Install, build, compile to any backend, run your first evaluation
GRDL Language Reference	Complete specification of the rule YAML format
The Seven Governance Laws	The universal governance primitives behind CFAIS
CFAIS Engine Architecture	How the deterministic evaluator works internally
Backends	OpenShell, Docker, and Standalone — how rules compile to each runtime
Agent Integration Guide	Connecting GRDL to Gemma 4, OpenClaw, LangChain, and custom agents
Governance Templates	Enterprise, DAO, and AI Safety templates explained
Writing Custom Rules	Step-by-step guide to authoring your own governance rules
CLI Reference	All CLI commands with examples
Sidecar API Reference	HTTP endpoints exposed by the CFAIS sidecar
Design Decisions	Why Go, why deterministic, why runtime-agnostic, why not ML
Patent and IP	Relationship between patents and open-source code
FAQ	Common questions

Quick links

• Repository: github.com/prahari-ai/grdl
• Install: go install github.com/prahari-ai/grdl/cmd/prahari@v0.2.0
• Go package: pkg.go.dev/github.com/prahari-ai/grdl
• License: Apache 2.0 (core) / BSL 1.1 (enterprise modules)
• Built by Prahari.ai