Hermes advanced delegation is an official, free plugin from Nous Research that lets Hermes Agent send each subagent task to its own model or its own Hermes profile.
Normally, every subagent in a Hermes batch runs on the same model, which means you either overpay on simple jobs or underpower hard ones.
The plugin adds two modes, called model routes and profile routes, and the agent picks a route per task from names and descriptions you write.
For a business, that is the difference between one generalist doing everything and a small team where each job goes to the right specialist.
In this post I'll explain what the plugin is and how to set it up, then walk through five ways I'd use it in an agency or business.
Hermes Advanced Delegation In One Table
This is the business summary before we get into the use cases.
| Question | Short answer |
|---|---|
| What is it? | It is an official plugin that routes each subagent task to a chosen model or Hermes profile. |
| Who makes it? | Nous Research, as part of the Hermes Official Plugins. |
| What does it cost? | The plugin is free, and you pay only for the models your routes use. |
| What does it need? | It needs Hermes 0.22.5 or newer. |
| What are the two modes? | Model routes pick a model per task, and profile routes pick a whole profile per task. |
| Can the agent see your keys? | No, because only route names and descriptions reach the model. |
| What's the main risk? | Profile routes share the listed profile's credentials and terminal access. |
🔥 Want my exact How I'd Use Hermes Advanced Delegation In A Business setup, step by step? Inside the AI Profit Boardroom you get step-by-step video trainings, my installable Agent OS and four coaching calls a week with 3,400+ members building real automations. → Join the AI Profit Boardroom
What Hermes Advanced Delegation Is
Hermes Agent's main agent can hand work to subagents through a tool called delegate_task.
By default, every one of those subagents runs on the same model, set by delegation.model or inherited from the main agent.
The advanced-delegation plugin adds two more modes next to that default.
Model routes let you define a short list of named options, such as fast, deep and local, each with its own provider and model.
Profile routes let a subagent run as one of your existing Hermes profiles, with that profile's own model, credentials, skills and SOUL.md.
The agent picks one route per task by reading the names and one-line descriptions you wrote.
It never sees the raw provider, model id, endpoint or key, so it can't invent a route that you didn't approve.
If you want the background on subagents themselves, my sibling post on Hermes Agent subagents covers it.
Why Routing Matters For Your Costs
Nous's delegation docs say a parallel batch of subagents typically burns the large majority of a run's tokens.
The built-in default gives you one model for the whole batch.
So if one task in ten needs a strong model, the other nine pay for it too, or the hard task suffers on a cheap model.
Model routes fix that, because each task picks the route that fits it.
I won't promise a saving figure, because it depends on your mix of tasks and your providers' pricing.
How To Set It Up
The setup is short, and every command here comes from the plugin's own page.
- Update to Hermes 0.22.5 or newer with
hermes update, because older versions refuse the plugin. - Install it in the profile that delegates with
hermes plugins install advanced-delegation. - Run
hermes delegation config, choose Model routes or Profile routes and build your list. - Run
hermes delegation config --showto check what you saved. - Start a fresh chat with
/new, because mode and route changes apply to new sessions.
Desktop users can install it from Settings → Model → Subagents → Install plugin instead.
Each route needs a name, a one-line description, a model and a provider or endpoint.
You can add a reasoning effort and, where the route offers one, a speed tier of normal, fast or ultrafast.
🔥 Want my Hermes agent setups for your business? Inside the AI Profit Boardroom, I've got Hermes Agent training, the Agent OS and step-by-step walkthroughs on profiles, models and subagents. You'll be learning alongside 3,400+ members with four coaching calls every week. → Get access here
Use Case 1: An Agency Mixing Cheap Research With Deep Strategy
Picture an SEO agency agent asked to audit five competitors and then recommend a strategy.
The five audits are fetch-and-summarise jobs, and the recommendation needs careful reasoning.
With model routes, I'd set up two options, using the plugin page's own example as a starting point.
delegation:
backend: model_map
backends:
model_map:
options:
- name: fast
description: "Cheap and quick: lookups, summaries, simple edits."
provider: openrouter
model: google/gemini-3-flash-preview
- name: deep
description: "Slow and careful: hard debugging and design work."
provider: anthropic
model: claude-opus-5
reasoning_effort: high
default: fast
The audits pick fast, and any task that needs deep reasoning picks deep.
Every result reports which route, provider and model it actually ran on, so you can check the agent's choices.
Use Case 2: A Business That Keeps Client Data Private
Some client files should never go to a cloud model.
Model routes accept a direct endpoint, and the plugin page shows a local route that points at a server on your own machine.
- name: local
description: "Private: runs on this machine."
base_url: http://localhost:8080/v1
model: qwen3-coder
Describe it clearly, such as "Private: runs on this machine", so the agent knows when to choose it.
Tasks that touch private data can then run locally while public research still uses a cloud route.
Use Case 3: A Team Of Specialist Profiles
Profile routes suit a business that has already built specialist Hermes profiles.
The plugin page's example lists a coder profile that writes and refactors code and runs tests, and a researcher profile for web research and long-form summaries.
When the main agent delegates, it hands coding tasks to the coder profile and research tasks to the researcher profile.
Each subagent runs exactly as if you'd started hermes -p <profile> for that task, with that profile's own skills and model.
That means the skills you've already built for each role get reused automatically.
🚀 Ready to put How I'd Use Hermes Advanced Delegation In A Business to work in your business? Inside the AI Profit Boardroom you get step-by-step video trainings, my installable Agent OS and four coaching calls a week with 3,400+ members building real automations. → Join the AI Profit Boardroom
Use Case 4: Watching And Steering Routed Work
My video at the top shows how I watch subagents live and steer or stop them mid-task, with three builds: a roadmap page, an SEO ROI calculator and an audit landing page.
The plugin page says Hermes core keeps the whole subagent lifecycle, including background runs and results, while the plugin only decides the route.
So the same habits apply when routes are switched on.
Watch the first routed batch live, check which route each task chose and steer early if a task picked the wrong one.
If the agent keeps choosing badly, rewrite the route descriptions, because they're the only thing it reads.
Use Case 5: Switching Modes For Different Kinds Of Day
Some days are research-heavy and some are build-heavy, and the plugin makes switching quick.
Typing /delegation in a chat shows your current mode and a picker, with buttons on Telegram, Discord and Slack.
/delegation model_map saves model routes for the next session, and /delegation default puts you back on the built-in behaviour.
Adding --now applies the change to the current chat, at the cost of one prompt-cache break, which the plugin page explains makes the next request re-read the conversation at full price.
Each mode keeps its own settings, so switching back and forth doesn't lose your routes.
The Risks Every Business Should Know
The biggest one is the security warning on the plugin page.
Listing a profile in profile routes lets the delegating agent's subagents use that profile's credentials, credential pool, SOUL.md and terminal policy.
So only list profiles whose keys and terminal access you're happy to share.
The second is that failures are loud by design.
If the plugin is missing or disabled, delegate_task refuses with an error naming the fix rather than quietly using the default route.
A task that can't be routed is refused on its own, while the rest of the batch keeps going.
The third is fallbacks, because a model route without its own fallback_providers uses delegation.fallback_providers, and a profile route uses that profile's chain.
If routed subagents run shell commands, put them behind a sandbox, and my Hermes Agent sandbox guide shows how.
Model Routes Or Profile Routes For Your Business?
| Your situation | The mode I'd pick | Why |
|---|---|---|
| You use one profile and want cheaper bulk work. | Model routes | It needs nothing beyond a list of models. |
| You must keep some work on your own hardware. | Model routes with a local route | The local route keeps that work on your machine. |
| You already run specialist profiles. | Profile routes | Each task reuses that profile's skills and model. |
| You're still learning how your agent delegates. | Default | A single delegation.model pin is simpler to start with. |
For the single-pin option, see my Hermes subagent model post, and for model names see my best Hermes Agent model ranking.
If you'd like help working out where agents fit your SEO business, you can book a free SEO strategy session with my team.
🔥 Want the full Agent OS with Hermes profiles ready to route? The AI Profit Boardroom includes the installable Agent OS, Hermes training and four weekly coaching calls where you can ask me how to design routes for your business. → Join the AI Profit Boardroom
My video on the Hermes 0.22 release covers the release line this plugin builds on.
Related Reading
- Hermes Agent subagents shows how delegation works and how businesses use it.
- Hermes subagent model explains the simpler single-model option.
- Hermes Agent v0.22 covers the release line behind these features.
Also On Our Network
- 🌐 The five-step Hermes advanced delegation install guide
- 🌐 Model routes versus profile routes, with a verdict
- 🌐 The Agent OS quick start for Hermes advanced delegation
- 🌐 GoldieBench scores for the models worth routing to
FAQ: Hermes Advanced Delegation
What is Hermes advanced delegation?
It is an official Nous Research plugin that adds model routes and profile routes to delegate_task.
Is the Hermes advanced delegation plugin free?
Yes, the plugin is free, and you pay only for the model providers your routes use.
How does advanced delegation help a business?
It sends simple tasks to cheaper models and hard tasks to stronger ones in the same batch, and it can hand work to specialist profiles.
Is it safe to use profile routes with client work?
Only list profiles whose keys and terminal access you're happy to share, because listed profiles' credentials become usable by the delegating agent.
What version of Hermes do I need?
You need Hermes 0.22.5 or newer.
If you run an agency and want a plan for putting agents on client work, you can book a free strategy session and we'll map it out.
📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉
🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉
About Julian
I'm Julian Goldie, an SEO entrepreneur, author and founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
- I've built a 7-figure SEO and link-building agency, Goldie Agency, with 70+ people in Bangkok.
- I've grown a YouTube channel to 400,000+ subscribers.
- I wrote the book "Link Building Mastery", which covers nine link building strategies.
- I run GoldieBench, where I score AI models on real one-shot builds.
→ Get my best AI training inside the AI Profit Boardroom
Start with two model routes, write clear descriptions and watch the first batch live, and Hermes advanced delegation will put every task on the right brain for your business.
