Skip to content

Hello World Template

Harness Engineering template: this showcase demonstrates one concrete capability slice of orchestrator as a control plane for agent-first software delivery.

Purpose: The simplest runnable workflow — one Workspace, one Agent, one Workflow. Zero API cost.

Use Cases

  • First contact with orchestrator — verify installation and basic flow
  • Understand the Workspace → Agent → Workflow resource relationship
  • Starting skeleton for custom workflows

Prerequisites

  • orchestratord is running (orchestratord --foreground --workers 2)
  • Database initialized (orchestrator init)

Steps

1. Deploy Resources

bash
orchestrator apply -f docs/workflow/hello-world.yaml --project hello-world

2. Verify Resources

bash
orchestrator get workspaces --project hello-world
orchestrator get agents --project hello-world
orchestrator get workflows --project hello-world

3. Create and Run a Task

bash
orchestrator task create \
  --name "hello" \
  --goal "Say hello" \
  --workflow hello \
  --project hello-world

4. Inspect Results

bash
orchestrator task list --project hello-world
orchestrator task info <task_id>
orchestrator task logs <task_id>

Expected Output

The echo agent returns a fixed JSON response:

json
{
  "confidence": 0.95,
  "quality_score": 0.9,
  "artifacts": [{
    "kind": "analysis",
    "findings": [{
      "title": "hello-world",
      "description": "Workflow executed successfully.",
      "severity": "info"
    }]
  }]
}

The task completes within seconds with status Completed.

Customization Guide

Replace with a Real Agent

Swap the echo agent's command for a real AI agent:

yaml
# Claude Code
command: claude -p "{prompt}" --verbose --output-format stream-json

# OpenCode
command: opencode -p "{prompt}"

You will need to configure the corresponding API key (via SecretStore or environment variables).

Add More Steps

Add steps to the Workflow's steps list and ensure the Agent has the matching capability.

Further Reading