agentos.guide › blog
A Hermes Agent n8n integration gives your Agent OS workspace two new powers: the agent can run and build n8n workflows through the official MCP entry, and n8n can hand work to the agent through webhook routes or the API server.
Book a free SEO strategy session →
This walkthrough sets up all three inside a typical Agent OS install, with the files to touch and the checks to run.
If you want agent workflows like this pointed at search growth, book a free SEO strategy session.
Source docs: Hermes MCP, webhooks, API server and n8n MCP.
In n8n, open Settings, then Instance-level MCP, and turn it on.
Open Connect and copy the Server URL. It ends in /mcp-server/http.
On older n8n versions it appears on the MCP settings page instead.
hermes mcp install n8n-official
Paste the Server URL when asked, then finish the OAuth approval in your browser.
If the browser step fails or times out, run it again:
hermes mcp login n8n-official
Hermes probes the server and shows the tools it found.
Keep the search and run tools on, and switch off anything that edits or deletes workflows until you trust the setup.
hermes mcp configure n8n-official
Then start a new session or type /reload-mcp in an open one.
Ask the agent: list my n8n workflows that are available to you.
If the list is empty, mark at least one workflow as available in MCP inside n8n and ask again.
Then ask it to run a harmless workflow, such as a test that writes one row to a sheet, and confirm the row appears.
n8n documents four groups of actions for MCP clients: searching workflows, interacting with workflows exposed to MCP, triggering and testing them, and from n8n 2.13.0 creating and editing workflows and data tables.
For a tightly scoped agent, n8n's MCP Server Trigger node exposes the tools of a single workflow instead of the whole instance.
Add to ~/.hermes/.env:
Book a free SEO strategy session →
WEBHOOK_ENABLED=true
WEBHOOK_PORT=8644
WEBHOOK_SECRET=replace-with-a-long-random-string
Restart the gateway and check health:
curl http://localhost:8644/health
Routes sit under platforms, webhook, extra, routes.
platforms:
webhook:
enabled: true
extra:
routes:
n8n-lead:
secret: "route-specific-secret"
prompt: "New lead from {{name}}: {{message}}. Draft a short reply for approval."
skills: []
deliver: "telegram"
You can also create routes from the command line with hermes webhook subscribe.
Payload fields are untrusted, so keep the prompt asking for a draft, not an action.
Add an HTTP Request node with method POST and URL http://your-hermes-host:8644/webhooks/n8n-lead.
Send the lead fields as JSON in the body.
For production, add a Code node before it that computes the generic signature and timestamp headers from the route secret.
The agent's draft arrives in the Telegram home channel set for your workspace.
| Option | Use in Agent OS |
|---|---|
| filters | Drop events that do not need the agent |
| script | Clean or reshape the payload with a file in ~/.hermes/scripts/ |
| deliver_only | Forward a rendered notification with no model call |
| cron_job | Fire an existing scheduled job instead of a new session |
| coalesce | Merge rapid updates about the same item |
| toolsets | Limit what this route's runs can touch |
🔥 Want the Agent OS build files for this?
Inside the AI Profit Boardroom you get step-by-step video trainings, the installable Agent OS, four coaching calls a week and 3,400+ members wiring agents into their tools.
Add to ~/.hermes/.env:
API_SERVER_ENABLED=true
API_SERVER_KEY=replace-with-a-long-random-key
Restart the gateway; the server listens on http://127.0.0.1:8642.
curl http://localhost:8642/v1/chat/completions \
-H "Authorization: Bearer replace-with-a-long-random-key" \
-H "Content-Type: application/json" \
-d '{"model": "hermes-agent", "messages": [{"role": "user", "content": "Reply with OK"}]}'
Use an HTTP Request node with POST, the same URL, a Bearer header and the JSON body above, with your real prompt in the content field.
Book a free SEO strategy session →
The response's message content is the agent's answer, which the next node can read.
For multi-step conversations, use /v1/runs or /v1/responses with an X-Hermes-Session-Key header so context carries between calls.
| Agent OS job | Part |
|---|---|
| "Run the client report workflow and summarise the result" | Part 1, MCP |
| "When a form fires, draft a reply and send it to my Telegram" | Part 2, webhook route |
| "Label this support ticket and return the label to n8n" | Part 3, API server |
| "Each morning, check last night's n8n failures" | Part 1 plus a Hermes cron job |
📞 Workspace wired. Now aim it at growth.
Book a free SEO strategy session and my team will show you which SEO and AI search jobs are worth automating first.
Every agent run started by a route or the API server shows up in your Hermes session history, so you can review what the agent did and why.
Tag routes by purpose in their names, such as n8n-lead or n8n-support, so history searches stay clean.
On the n8n side, keep execution logging on for the workflows the agent can run, so both halves leave a trail.
Store the webhook secret and API key in ~/.hermes/.env, never inside a route prompt or an n8n note.
Book a free SEO strategy session →
In n8n, keep the secret in a credential rather than pasting it into a Code node in plain text.
Rotate both after anyone leaves the team.
If you run separate Hermes profiles for different clients, install the MCP entry in each profile that needs n8n access and give each its own routes.
Keep one gateway per profile so the webhook and API ports do not clash.
The older third-party n8n bridge is no longer installable from the catalog.
If your workspace already has it, it keeps working as a custom MCP server alongside the official entry.
Copy patterns from the automation blueprints and the MCP guide.
The n8n Webhook node docs help if you deliver agent results back into n8n.
Members of the AI Profit Boardroom post their Agent OS and n8n builds every week.
Book a free SEO strategy session →
Install the n8n-official MCP entry with the Server URL from n8n Instance-level MCP, approve OAuth, and choose the tools to keep.
Yes. Enable the webhook adapter, add a signed route in config.yaml and POST to it from an HTTP Request node.
Call the API server at /v1/chat/completions on port 8642 with a Bearer key and model hermes-agent.
n8n only exposes workflows marked available in MCP. Enable at least one and ask again.
Set up Part 1 today, run one harmless workflow from chat, then add a route or the API server when a flow needs the agent.
For growth on top, book a free SEO strategy session, and keep this Hermes Agent n8n integration guide in your Agent OS notes.
About Julian
I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.
→ Hermes Agent Cron Job: Schedule, Stop And Recover Jobs In Your Agent OS With 5 Copy-Paste Prompts
→ Hermes Agent Home Assistant: Setup + 3 Agent OS Prompts
→ Hermes Agent Gmail: Setup + 3 Inbox Prompts for Agent OS
The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.
Join the AI Profit Boardroom →📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉