Agents are infrastructure.

Orloj is the full stack for agentic systems.

Build, run, govern, and observe multi-agent systems from one declarative stack: agents, tools, models, memory, schedules, webhooks, approvals, policies, workers, traces, metrics, and deployment primitives.

Quickstart | Docs | Stack | How It Works | Screenshots

Status: Orloj is under active development. APIs and resource schemas may change before 1.0.

What Is Orloj?

Orloj is the open-source full stack for developing and operating agentic systems.

An agent system is no longer just a prompt and a loop. It has model routing, tool permissions, credentials, memory, retries, human approvals, schedules, webhooks, workers, traces, logs, metrics, and deployment topology. Orloj treats those pieces the way infrastructure platforms treat services: as versioned resources with desired state, status, controllers, leases, workers, runtime policy, and operational visibility.

The Orloj Agent Stack

Orloj is not a single runtime component. It spans the layers teams need to develop, run, govern, and observe agentic systems.

How The Stack Operates

Orloj Is Right For You If

Problems Orloj Solves

What Is In The Stack

Example: An Agent System As Code

Define an agent:

apiVersion: orloj.dev/v1
kind: Agent
metadata:
  name: research-agent
spec:
  model_ref: openai-default
  prompt: |
    You are the research stage.
    Produce concise, verifiable findings for the writer.
  tools:
    - web_search
  allowed_tools:
    - web_search
  limits:
    max_steps: 6
    timeout: 30s

Bind it to a model endpoint:

apiVersion: orloj.dev/v1
kind: ModelEndpoint
metadata:
  name: openai-default
spec:
  provider: openai
  base_url: https://api.openai.com/v1
  default_model: gpt-4o
  auth:
    secretRef: openai-api-key

Wire agents into a graph:

apiVersion: orloj.dev/v1
kind: AgentSystem
metadata:
  name: report-system
spec:
  agents:
    - planner-agent
    - research-agent
    - writer-agent
  graph:
    planner-agent:
      edges:
        - to: research-agent
    research-agent:
      edges:
        - to: writer-agent

Run it as a task:

apiVersion: orloj.dev/v1
kind: Task
metadata:
  name: weekly-report
spec:
  system: report-system
  input:
    topic: enterprise AI copilots
  retry:
    max_attempts: 2
    backoff: 2s
  message_retry:
    max_attempts: 2
    backoff: 250ms
    max_backoff: 2s
    jitter: full

Governance Built Into The Runtime

Governance in Orloj is not a documentation convention. It is evaluated while work is being executed.

Unauthorized actions fail closed and appear in the task trace.

How It Works