Aurora: The Hermes Swarm Orchestrator Explained

A Hermes Swarm without Aurora is just a pile of agents — Aurora is the orchestrator that turns them into an actual team. After running her on real missions for months, here's exactly what she does and how to talk to her.

This post covers Aurora specifically. What she does, how she routes missions, how to talk to her, and how she coordinates the rest of your swarm.

What Aurora Is

Aurora is the main agent at the top of your Hermes Swarm — the orchestrator. You don't talk to sub-agents directly. You talk to Aurora and she handles the rest.

What Aurora Does

Three jobs.

1 — Receives your mission

You describe what you want in plain English and Aurora ingests it.

2 — Composes a routing plan

Aurora figures out which sub-agents are needed, what each should do, what order they should work in, and how they should hand off.

3 — Coordinates execution

Aurora deploys sub-agents, tracks progress, handles handoffs between them, and reports back to you.

Why Aurora Matters

Without an orchestrator you'd manually assign work to each sub-agent, track progress yourself, handle handoffs, and become the bottleneck of your own system.

With Aurora you set the goal and she handles the rest. That's the leverage of multi-agent.

How To Talk To Aurora

Plain English missions work best.

A bad mission looks like "Use Builder agent to write blog post then have Reviewer check it" — too prescriptive, since Aurora can figure that out herself.

A good mission looks like "Create a 2,000-word blog post on AI SEO, optimised for ranking, with internal links and FAQ" — high-level goal that lets Aurora plan the steps.

Trust Aurora to plan.

🔥 Want my full Aurora prompt playbook? Inside the AI Profit Boardroom, I share my exact Aurora prompts for content, customer ops, research, and more. Plus 2-hour Hermes course and weekly live coaching. 2,800+ members. → Get the playbook

What Makes Aurora Different From A Regular AI Agent

Three differences matter.

1 — She reasons about coordination

Regular agents focus on doing one task. Aurora focuses on orchestrating multiple tasks. That's a different cognitive load.

2 — She manages sub-agents

Aurora knows what each role can do (Builder, Reviewer, Triage, Lab, Sage, Scribeex) and matches mission steps to capabilities.

3 — She maintains mission-level state

Sub-agents focus on their step while Aurora maintains the bigger picture. The result is cohesive multi-agent output rather than disjointed pieces.

When Aurora Decides Single Agent vs Team

Aurora chooses one of two modes. Single agent mode is when the mission is small enough for one agent. Automatic team mode is when the mission needs multi-agent collaboration.

You can override in settings. For most missions, automatic team mode is the right choice.

Common Aurora Setup Mistakes

Four mistakes worth avoiding.

Too vague missions. "Help me with marketing" gives Aurora nothing to plan. Be specific about goal and scope.

Too prescriptive missions. "Use Agent X with Y prompt" defeats the orchestration. Let Aurora plan.

No success criteria. Tell Aurora what "done" looks like or missions drift.

Ignoring blocked agents. If a sub-agent is blocked, Aurora can't progress. Check sub-agent settings (usually a wrong API key).

Mission Templates That Work With Aurora

For different workflows.

Content mission

"Create a 2,000-word blog post on [topic]. Optimise for SEO. Include H2/H3 structure, FAQ, internal links to my existing content. Use my Hormozi-style tone."

Research mission

"Research [topic]. Pull from at least 5 sources. Synthesise findings. Output as a 1,000-word brief with citations."

Customer ops mission

"Triage my Telegram messages. Reply to common FAQs. Escalate complex issues to me. Maintain log of all interactions."

Code mission

"Refactor [module] to improve test coverage. Add tests where missing. Don't break existing functionality."

For each, Aurora plans the steps.

Working With Aurora's Handoffs

Aurora orchestrates handoffs between sub-agents and you can see this in the office view — round table or grid layout, agents passing work to each other, status updates per agent.

If a handoff fails, check sub-agent settings, then restart Aurora and re-route the mission.

Aurora Vs Single-Agent Hermes

Quick comparison. Single-agent Hermes is one agent doing one task with you providing instructions and linear execution. Aurora-led Swarm is multiple agents on a coordinated mission with you providing the goal and parallel plus sequential execution as needed.

For complex work, Aurora wins. For simple chat, single-agent is fine.

I cover this in detail in Hermes Swarm vs Single-Agent.

Aurora's Strengths

Where Aurora excels — multi-step missions, parallel-friendly work, quality-critical workflows with Reviewer agents, and handoff-heavy processes.

Aurora's Limits

Be honest. She doesn't reason at superhuman level, can misroute on edge cases, and sometimes needs guidance on truly novel missions.

Treat her as a competent project manager, not a genius strategist.

Phone Access For Aurora

Hermes Workspace supports phone access. You can trigger Aurora missions from your phone, check progress remotely, and approve actions on the go.

Effectively you're running a swarm operation from your pocket.

Daily Reality Of Working With Aurora

What it looks like. At 8am I give Aurora today's mission via desktop or phone. At 8:01 Aurora deploys sub-agents. Throughout the day sub-agents work in parallel while Aurora updates progress. I approve outputs as they're ready.

You're managing a team without managing humans.

Why Aurora Specifically (Not Just Any Orchestrator)

Aurora is built into Hermes Workspace's swarm plugin. She's free, open source, tightly integrated with Hermes capabilities, and designed for the Hermes role system (Builder, Reviewer, etc.).

For the Hermes ecosystem, Aurora is the obvious pick. For other ecosystems, similar orchestrators exist (e.g. Aion CLI for OpenClaw plus Aion UI).

🚀 Want my full Aurora workflow playbook? The AI Profit Boardroom has my Aurora mission templates, 2-hour Hermes course, OpenClaw 6-hour course, daily training, weekly live coaching. 2,800+ members. → Join here

FAQ — Aurora Hermes Swarm Orchestrator

Can I rename Aurora?

In settings, yes — but the role stays the same.

What if Aurora misroutes a mission?

Refine the mission prompt and be more specific about the goal.

Does Aurora work without sub-agents?

She defaults to single-agent mode if you have no sub-agents added.

Can I run multiple Aurora instances?

Yes — each in a separate workspace.

Does Aurora learn from past missions?

Limited — she has memory but not deep self-improvement (unlike Auto Research Claw).

Should I always use Aurora?

For multi-step missions, yes. For trivial chat, single-agent is fine.

Can I customise Aurora's prompts?

Yes — like any sub-agent, her system prompt is customisable.