Hermes Multiple Profiles: One Gateway for Every Hermes Agent

Hermes multiple profiles let you run several separate Hermes agents on one computer — say a coding agent, a research agent and a customer support agent — each with its own settings, memory and bots. With Hermes Agent v0.22, all of those profiles now run through one messaging gateway and one desktop backend, instead of a separate process for each. Create a new profile and its bots go live without a restart. In this guide I explain what a profile is, how the new shared gateway works, what stays separate, and how to set it up. Watch the video below, then follow the steps.

What is a Hermes profile?

According to the Hermes profiles docs, a profile is a separate Hermes home. Each one has its own config.yaml, .env file with keys, SOUL.md personality, memories, sessions, skills, cron jobs and database. Profiles live under ~/.hermes/profiles/.

In plain English: each profile is its own agent. Your coding agent doesn’t share memory with your support agent, and your support bot doesn’t have your personal API keys. That separation is the whole point.

Creating one is a single command: hermes profile create coder. Hermes also creates a coder command for it, so you can run coder setup and coder chat straight away.

What changed in v0.22: one gateway for all profiles

Before Hermes Agent v0.22.0, each profile that talked to Telegram, Discord, Slack and so on needed its own gateway process. Run five profiles and you had five gateways to start, watch and restart.

Now, as the multi-profile gateways guide explains, one messaging gateway and one desktop backend per machine serve every profile. Nous calls this multiplexing, and it’s on by default. When you create a profile, its bots go live without a restart.

Each profile still keeps its own credentials, allowlists, cron jobs and sessions. Only the process is shared. Upgrading is painless too: hermes update migrates per-profile setups to the new model automatically.

What stays separate between profiles

Sharing one process could sound risky, so it’s worth knowing what Hermes keeps apart. The docs list a long table of things each profile resolves from its own files. The ones that matter most:

• Keys and bot tokens come from each profile’s own .env. A profile never falls back to the default profile’s keys.
• Who can talk to the bot (allowlists) is set per profile, and closed by default.
• Sessions are stored per profile, so two profiles in the same chat never share history.
• Command approvals are per profile — an “always allow” in one profile doesn’t approve anything in another.
• Logs sit under each profile’s own home.

What’s shared by design is the process itself, its lock and status file, and the one HTTP listener. Webhooks for each profile are served under /p/<profile>/ on that shared listener.

One rule to remember for chat platforms: each profile needs its own bot token. If two profiles use the same Telegram token, Hermes parks the duplicate and keeps everything else running.

How to set up multiple Hermes profiles

• Create a profile: hermes profile create coder. Add --description to label it, or --clone to copy settings from your current profile.
• Set it up: coder setup, then coder chat to talk to it.
• See them all: hermes profile list.
• Switch your default: hermes profile use coder, and hermes profile use default to go back.
• Check the gateway: hermes gateway status shows what’s running and explains any refusals.

Cloning is careful by default. Clones skip bot tokens unless you add --clone-channels, and profile exports skip your .ssh, .aws and .gnupg folders.

In Hermes Desktop, you can open a profile in its own window or set it as the startup default.

Managing profiles on the shared gateway

Because there’s now one gateway, the per-profile commands work a little differently:

• hermes -p coder gateway stop parks just that profile; start or restart brings it back. The rest keep running.
• hermes gateway migrate --multiplex migrates by hand if you need to, with --dry-run to preview first.
• gateway.standalone: true in a named profile’s config is a temporary opt-out if one profile must run its own gateway.
• Named profiles refuse gateway install and start without --force, so you don’t accidentally start a second gateway.

Setting gateway.multiplex_profiles to false is retired, so the shared gateway is the way forward.

3 ways to use multiple profiles

• One agent per job. A coding profile with developer tools, a research profile with web access, and a support profile connected to your customer chat. Each has the right tools and nothing more.
• One agent per client. Agencies can keep each client’s keys, bots and memory in a separate profile, so nothing leaks between accounts.
• Test before production. Try new skills, plugins or models in a test profile before touching the agent your team relies on.

Profiles pair well with delegation. If you want one agent to hand work to others, read my guides to Hermes advanced delegation and choosing a Hermes subagent model.

Safety notes

Two warnings from the docs are worth repeating. First, never point two agents at the same profile — give each its own. Second, profiles separate settings and data, but they are not a sandbox. A profile running commands on your machine can still reach your files, so set terminal.cwd explicitly and use a sandboxed terminal backend for anything risky.

My verdict on Hermes multiple profiles

Running several agents used to mean juggling several gateways. Now one process serves them all, new profiles go live instantly, and each one still keeps its own keys, bots and memory. If you’ve been cramming every job into one agent, Hermes multiple profiles are the cleaner way to grow into a small team of agents.

About Julian Goldie

I’m Julian Goldie — founder of Goldie Agency and the AI Profit Boardroom. I test new AI agents almost every day for 432K+ YouTube subscribers and run Hermes, OpenClaw, OpenCode, Claude and Codex side by side in my own Agent OS. Claim your free strategy session or join the AI Profit Boardroom.

Want another angle? Read the AI Profit Boardroom guide to Hermes multiple profiles.

FAQ: hermes multiple profiles

What are Hermes profiles?

A profile is a separate Hermes home with its own config, keys, personality, memory, sessions, skills and cron jobs. Each profile acts as its own agent.

How do I create a new Hermes profile?

Run hermes profile create followed by a name, for example hermes profile create coder. Then run coder setup and coder chat.

Do multiple Hermes profiles need separate gateways?

Not any more. Since v0.22, one messaging gateway and one desktop backend per machine serve every profile, and new profiles go live without a restart.

Can two Hermes profiles use the same Telegram bot?

No. Each profile needs its own bot token. If two profiles use the same token, Hermes parks the duplicate and keeps the rest running.

Are Hermes profiles a security sandbox?

No. Profiles keep settings and data separate, but they don’t sandbox commands. Set terminal.cwd and use a sandboxed terminal backend for risky work.

Related reading

• Hermes Advanced Delegation
• Hermes Subagent Model
• Hermes Agent v0.22

Last updated October 2026. This page is a living guide to hermes multiple profiles — the facts here move fast and it is updated as they do.