# Orloj — Extended Reference > Orloj is an open-source orchestration plane for multi-agent AI systems. Define agents, tools, policies, and workflows as declarative YAML manifests. Orloj handles scheduling, execution, governance enforcement, and reliability so your agent systems operate with the same rigor as the rest of your infrastructure. **Site:** www.orloj.dev | **Docs:** docs.orloj.dev | **GitHub:** github.com/OrlojHQ/orloj --- ## Getting Started & Introduction - **[What is Orloj](https://www.orloj.dev/blog/what-is-orloj)** — Quick introduction to Orloj and agent orchestration as a category. Start here if you're new to the concept. Covers the basic problem: running AI agents in production today is ad-hoc and ungoverned, like containers before Kubernetes. - **[Introducing Orloj](https://www.orloj.dev/blog/introducing-orloj)** — The initial release announcement. Describes core features: declarative YAML manifests, built-in governance, fail-closed authorization, lease-based task ownership, WASM tool isolation, and server/worker architecture. - **[2026 Agent Orchestration Guide](https://www.orloj.dev/blog/2026-agent-orchestration-guide)** — Comprehensive market overview of agent orchestration landscape in 2026. Positions Orloj alongside other frameworks and explains why orchestration became a separate concern from agent building. --- ## Core Concepts & Architecture - **[What is Agent Orchestration](https://www.orloj.dev/blog/what-is-agent-orchestration)** — Defines agent orchestration and explains its role in production systems. Covers why frameworks alone (LangChain, CrewAI) are insufficient for production use, and what orchestration adds: scheduling, governance, observability, reliability patterns. - **[Declarative Agent Management](https://www.orloj.dev/blog/declarative-agent-management)** — Explains Orloj's YAML-first approach. Covers why declarative infrastructure is superior to imperative code for agent systems: version control, audit trails, immutability, policy enforcement. Includes manifest examples. - **[The Missing Layer: Orchestration Plane](https://www.orloj.dev/blog/missing-layer-orchestration-plane)** — Deep dive into why agent systems need a dedicated orchestration layer. Draws the Kubernetes analogy: containers need orchestration, agents need orchestration. Explains the three-layer architecture (agent framework, orchestration plane, governance layer). - **[Fail-Closed vs Fail-Open](https://www.orloj.dev/blog/fail-closed-vs-fail-open)** — Governance philosophy. Explains why Orloj defaults to denying unauthorized actions (fail-closed) rather than allowing them by default (fail-open). Covers security implications and operational consequences. --- ## Production Operations & Reliability - **[Agent Observability](https://www.orloj.dev/blog/agent-observability)** — Monitoring, tracing, and debugging multi-agent systems. Covers structured logging, distributed tracing, metrics collection, and how to troubleshoot agent behavior in production. Essential for operational visibility. - **[Agent Cost Attribution](https://www.orloj.dev/blog/agent-cost-attribution)** — Understanding and allocating costs across your agent fleet. Covers how to track API costs, compute costs, and model costs per agent, per workflow, per user. Critical for financial governance of agent systems. - **[Agent Failures at 3am](https://www.orloj.dev/blog/agent-failures-3am)** — Operational resilience patterns and common failure modes. Covers retry strategies, dead-letter handling, timeout management, idempotency tracking, and how Orloj prevents cascading failures in multi-agent systems. - **[Testing Agent Systems](https://www.orloj.dev/blog/testing-agent-systems)** — Strategies for validating multi-agent workflows. Covers unit testing agents, integration testing systems, mutation testing, and how to validate governance policies. Emphasizes deterministic and repeatable testing approaches. - **[Multi-Agent Systems Reliability](https://www.orloj.dev/blog/multi-agent-systems-reliability)** — Building systems that survive production. Covers lease-based task ownership, idempotency guarantees, workflow state machines, recovery patterns, and SLO design for multi-agent workloads. --- ## Governance & Compliance - **[Agent Governance Strategies for Enterprise](https://www.orloj.dev/blog/agent-governance-strategies-enterprise)** — Governance frameworks and policies for large organizations running agent systems. Covers role-based access control, tool permissions, approval workflows, audit logging, and policy as code. - **[The Governance Gap](https://www.orloj.dev/blog/governance-gap)** — Explains how governance became separated from orchestration. Why frameworks don't include governance, how organizations retrofit it, and why it should be built in from the start. - **[Healthcare Governance](https://www.orloj.dev/blog/healthcare-governance)** — Compliance patterns for HIPAA and regulated healthcare environments. Covers data handling, audit trails, approved tool constraints, and how Orloj enforces healthcare policies. - **[Financial Services Governance](https://www.orloj.dev/blog/financial-services-governance)** — Governance for fintech, banking, and payment systems. Covers transaction logging, non-repudiation, audit requirements, and how fail-closed governance prevents unauthorized transactions. - **[EU AI Act and Agent Systems](https://www.orloj.dev/blog/eu-ai-act-agent-systems)** — Regulatory considerations for deploying agent systems in Europe. Covers transparency requirements, audit trails, human oversight, and how Orloj's governance layer helps achieve compliance. - **[Guardian Agents](https://www.orloj.dev/blog/guardian-agents)** — Using agents to govern other agents. Covers hierarchical governance patterns where specialized agents monitor, validate, and constrain the actions of other agents in the system. --- ## Comparisons & Market Analysis - **[Orloj vs LangGraph vs CrewAI](https://www.orloj.dev/blog/orloj-vs-langgraph-vs-crewai)** — Head-to-head feature comparison of three popular multi-agent platforms. Breaks down when to use each: LangGraph for building agents, CrewAI for agent collaboration, Orloj for production operations. - **[Top 5 Agent Orchestration Platforms](https://www.orloj.dev/blog/top-5-agent-orchestration-platforms)** — Market overview of the agent orchestration category. Compares Orloj, LangGraph, CrewAI, AutoGen, and other platforms across dimensions: governance, observability, reliability, deployment models. - **[The Cost of Ungoverned Agents](https://www.orloj.dev/blog/cost-of-ungoverned-agents)** — Business and technical case for agent governance. Quantifies the costs of failed audits, security incidents, runaway model costs, and operational incidents caused by ungoverned agent behavior. --- ## Use Cases - **[Healthcare Agent Governance](https://www.orloj.dev/use-cases/healthcare)** — How healthcare organizations use Orloj for HIPAA-compliant AI agent orchestration. Covers PHI access control, human-in-the-loop enforcement for clinical decisions, audit trail requirements, and declarative governance with AgentPolicy, AgentRole, and ToolPermission resources. - **[Financial Services Agent Governance](https://www.orloj.dev/use-cases/financial-services)** — How financial services firms use Orloj for SEC, OCC, and FINRA-compliant agent orchestration. Covers model risk management (SR 11-7), trading approval workflows via ToolApproval, audit trails for regulatory examination, and human oversight enforcement. ## Comparisons - **[Orloj vs LangGraph](https://www.orloj.dev/compare/orloj-vs-langgraph)** — In-depth comparison of Orloj and LangGraph. Covers architecture differences, governance capabilities, orchestration patterns, and when to use each. - **[Orloj vs CrewAI](https://www.orloj.dev/compare/orloj-vs-crewai)** — In-depth comparison of Orloj and CrewAI. Covers multi-agent coordination, governance models, production readiness, and when to use each. --- ## Documentation & Quick Reference The full documentation is available at [docs.orloj.dev](https://docs.orloj.dev). Key sections: - **Concepts** — Core terminology: agents, systems, tools, workflows, policies, resources, manifests - **Architecture** — Three-layer design, server/worker model, WASM isolation, lease-based ownership - **Getting Started** — Installation, quickstart manifest, running your first agent - **Governance** — Policy definition, role-based access control, tool permissions, approval workflows - **Operations** — Monitoring, logging, troubleshooting, disaster recovery, multi-tenancy - **Roadmap** — Planned features, current phase, contribution guidelines --- ## About This Reference This file (llms-full.txt) is designed to help AI search engines (ChatGPT, Claude, Perplexity, Gemini) understand Orloj's content, discover relevant pages, and cite the source correctly. Each entry includes what the page covers so AI systems can determine relevance without fetching the full page. Updated: April 2026