Hermenix
Hermes Agent in a NixOS VM that owns its own bootloader and its own
/nix/store, on two btrfs-backed disks, under an unprivileged hardened QEMU.
THE ULTIMATE WAY TO RUN HERMES!
The agent can rebuild its own system and update itself, durably. It shares no filesystem with the host, and reaches nothing on your network unless you say so.
One CLI, hermenix, with five subcommands:
| For | |
|---|---|
hermenix setup |
one-off host provisioning — disks, cache key, image build |
hermenix migrate <src> |
carry an existing flat HERMES_HOME into the guest |
hermenix remote <host> "<cmd>" |
run setup/migrate on a remote host over SSH |
hermenix vm <sub> |
control the VM itself — start/stop/console/snapshots/backups |
hermenix ctl <sub> |
drive the agent inside the VM — identity/config/secrets/gateway |
Installation
Guided
The hermenix CLI (nu/hermenix.nu) is a flake package — no checkout needed
anywhere. Run it straight from its Gitea URL:
nix run git+ssh://gitea@minipc1.lab.internal:22/ari/hermenix.git#hermenix -- setup
(Or nix profile install git+ssh://gitea@minipc1.lab.internal:22/ari/hermenix.git#hermenix
once, then just hermenix setup from here on — or get it declaratively via
the nixos-setup overlay, same as any other package. Any of these work from
any directory, on any host with SSH access to Gitea.)
Walks the host-side steps: checks prerequisites, creates the disks, generates
the cache signing key, writes /etc/hermenix/local.nix from your SSH keys,
builds the bootstrap image and writes it to the disk. It prints the module
snippet to paste into your own configuration at the end, with your uplink
interface already filled in.
Safe to re-run — every step detects what's already done and offers to skip, and nothing destructive happens without an explicit confirmation.
It deliberately stops short of nixos-rebuild, since that's your config.
The rest of this section is what the command does, for when you'd rather do it by hand or something goes sideways.
1. Create the disks
Two of them, and the split is the whole design:
| Holds | Size | Snapshotted | Backed up | |
|---|---|---|---|---|
| system | NixOS, /nix/store |
80G | No | No |
| state | identity, memories, workspace, /etc/nixos |
20G | Yes | Yes |
The system disk is large, takes all the write churn, and is entirely rebuildable from this flake — so there's no point protecting it. The state disk holds the only things that can't be reproduced, which makes snapshots cheap and backups quick.
sudo btrfs subvolume create /var/lib/hermes-vm/system
sudo btrfs subvolume create /var/lib/hermes-vm/state
sudo mkdir -p /var/lib/hermes-vm/snapshots
# BEFORE creating any file inside them
sudo chattr +C /var/lib/hermes-vm/system /var/lib/hermes-vm/state
sudo btrfs property set /var/lib/hermes-vm/system compression none
sudo btrfs property set /var/lib/hermes-vm/state compression none
sudo truncate -s 80G /var/lib/hermes-vm/system/system.raw
sudo truncate -s 20G /var/lib/hermes-vm/state/state.raw
sudo mkfs.ext4 -L hermes-state /var/lib/hermes-vm/state/state.raw
The chattr +C ordering matters. Btrfs is copy-on-write and so is a VM
writing to a disk image; stacked, every guest write forks an extent and the
image fragments badly. +C turns that off — but only for files created
afterwards. Set it on empty subvolumes and confirm with lsattr before
writing anything. If you get it backwards, recreate the subvolume and copy
the image in fresh.
Separate subvolumes because a btrfs snapshot covers a whole subvolume. Keeping the system image out of the state subvolume is what makes snapshots small.
2. Generate the cache signing key
The host serves its store to the guest as a read-only binary cache, so the guest doesn't re-download from the internet what you already have.
sudo mkdir -p /var/lib/secrets
sudo nix-store --generate-binary-cache-key hermes-cache-1 \
/var/lib/secrets/hermes-cache.secret /var/lib/secrets/hermes-cache.pub
sudo chown nix-serve /var/lib/secrets/hermes-cache.secret
sudo chmod 0400 /var/lib/secrets/hermes-cache.secret
cat /var/lib/secrets/hermes-cache.pub
3. Write /etc/hermenix/local.nix
Host-specific settings live here, not in this repo — see
guest/local.nix.example for the reference shape:
{
cachePublicKey = "hermes-cache-1:AbCd..."; # the key you just printed; leave
# empty and the guest just skips
# the host cache
authorizedKeys = [ "ssh-ed25519 AAAA... you@laptop" ];
}
sudo mkdir -p /etc/hermenix first if it doesn't exist.
4. Build the image and write it to the system disk
The two extra substituter options stop you compiling Determinate Nix from
source. --impure is needed because guest/common.nix reads
/etc/hermenix/local.nix from outside this flake's own source tree —
deliberately, so the flake itself stays generic and shareable while your
settings stay on your host.
nix build --impure git+ssh://gitea@minipc1.lab.internal:22/ari/hermenix.git#hermes-guest-image \
--option extra-substituters https://install.determinate.systems \
--option extra-trusted-public-keys cache.flakehub.com-3:hJuILl5sVK4iKm86JzgdXW12Y2Hwd5G07qKtHTOcDCM=
# notrunc keeps the disk at its full size; the image is smaller
sudo dd if=result/nixos.img of=/var/lib/hermes-vm/system/system.raw \
bs=4M conv=notrunc,fsync status=progress
(Building from a local checkout instead — e.g. while developing this flake —
works the same way with . in place of the git URL.)
5. Add the module to your system flake
{
inputs.hermenix.url = "git+ssh://gitea@minipc1.lab.internal:22/ari/hermenix.git";
outputs = { nixpkgs, hermenix, ... }: {
nixosConfigurations.myserver = nixpkgs.lib.nixosSystem {
system = "x86_64-linux";
modules = [ hermenix.nixosModules.default ./configuration.nix ];
};
};
}
services.hermenix = {
enable = true;
disk.baseDir = "/var/lib/hermes-vm";
network.externalInterface = "eno1"; # your uplink
# network.bridged = true; # guest takes its own LAN IP via DHCP;
# # every port it opens is then LAN-reachable
vcpu = 4;
mem = 8192;
};
sudo nixos-rebuild switch, then hermenix vm console to watch it boot.
6. First boot, inside the guest
hermenix vm ssh
# Confirm the disk labels match guest/hardware.nix BEFORE any rebuild
lsblk -f
# FlakeHub Cache login. Device-code flow: it prints a URL you approve in a
# browser anywhere, so the headless guest never needs one.
determinate-nixd login
# Pin the seeded flake
sudo nix flake lock /etc/nixos
The Hermes CLI installs itself automatically on first boot (systemd.services.hermes-install
in common.nix) — it runs unattended, so there's nothing to run for that
part. Give it a few minutes (it's compiling native Node modules); watch it
with journalctl --user -u hermes-install -f if you want to see it happen.
Once it's done:
hermes setup
hermes gateway install
These two need real input (model provider, API keys, messaging platform tokens), so they stay manual.
The identity seed is already in place — SOUL.md, a starter config.yaml, a
USER.md and a skill covering safe self-rebuilds. Back on the host, check
the gateway came up and that the unit name matches the default:
hermenix ctl units
hermenix ctl status
hermenix ctl soul
Check the labels in step 6 before the first rebuild.
guest/hardware.nixassumes the root partition is labellednixosand the ESPESP. If the image generator used different ones, fix that file first. Otherwise the agent's firstnixos-rebuild switchproduces a system that won't boot and you're restoring a snapshot on day one.
From here hermes update self-updates in place and
sudo nixos-rebuild switch --flake /etc/nixos#hermes is a durable change to
the machine. Neither needs anything from you.
Running it on a remote host
setup and migrate both act on whatever machine they run on — local
disks, local sudo, vsock into the local guest. There's no way around SSHing
into the target host first. hermenix remote does that SSH step for you:
hermenix remote fwdesk.lab.internal setup
hermenix remote fwdesk.lab.internal "setup --clean"
hermenix remote fwdesk.lab.internal "migrate /var/lib/hermes"
Quote the subcommand and its own flags/args as one string whenever
there's more than just the bare subcommand name — nu's own argument parser
treats any bare --flag-shaped word as an attempt to match a flag on
remote itself, so hermenix remote host setup --clean (unquoted) fails
before it ever reaches the SSH step.
It runs nix run <flake>#hermenix -- <the rest of the command> directly on
the remote host — no checkout involved, on either end. The remote host does
need its own SSH access to Gitea (the same access a git clone would need).
Override the flake URL with --flake if you're pointing at a fork or a
branch.
It connects with ssh -t, which allocates a real pty — required for the
interactive prompts in setup/migrate to work at all over the connection.
SSH into the guest
hermenix vm ssh # interactive shell
hermenix vm ssh "systemctl status" # run a command and exit
hermenix vm console # follow the VM's boot/journal output
ssh takes its remote command as a single string, not separate words — quote
anything past a single bare word (nu's own parser otherwise tries to match
--flag-shaped tokens against vm ssh's own flags, and errors on extra
positional words either way).
This goes over vsock, not the network. There is no listening SSH port
exposed anywhere, and the guest needs no inbound network path for you to
reach it. console is just journalctl -u hermes-vm -f on the host — a
read-only follow of the VM unit's journal, useful for watching it boot or
diagnosing why SSH won't come up, but it can't take input. There's currently
no interactive fallback for when the guest itself is unresponsive.
Under the hood ssh is ssh -o ProxyCommand='socat - VSOCK-CONNECT:<cid>:22'
(cid defaults to 77, services.hermenix.vsockCid), if you'd rather drive it
by hand.
The VM itself is a systemd unit: hermenix vm start/stop/restart control
it directly, same as systemctl {start,stop,restart} hermes-vm on the host.
Bootstrapping Hermes's identity
The seed lives in guest/hermes/:
guest/hermes/
home/
SOUL.md -> ~/.hermes/SOUL.md
config.yaml -> ~/.hermes/config.yaml
memories/USER.md -> ~/.hermes/memories/USER.md
skills/nixos-self-config/SKILL.md -> ~/.hermes/skills/...
workspace/
AGENTS.md -> ~/workspace/AGENTS.md
Edit these before building the image and the agent wakes up already knowing who it is, how you like to be spoken to, and how to rebuild its own system without bricking itself.
Seeding happens on first boot only, and this is deliberate. Hermes
rewrites SOUL.md and its memories as it learns — that's the product. If the
seed were re-applied on every rebuild it would silently delete everything the
agent had worked out about you. So the copy is guarded by a marker file and
never repeats.
Two consequences:
- Editing
guest/hermes/after first boot changes nothing by itself. Usehermenix ctl reseedwhen you actually want to overwrite. - The seed is the starting identity, not the current one. To see what
Hermes has made of itself,
hermenix ctl soul.
Machine facts vs identity
/etc/hermes/ENVIRONMENT.md is generated from guest/common.nix on every
rebuild and describes the machine: what's reachable, where the store is, what
persists. It's always accurate and there's no point editing it.
SOUL.md is identity and belongs to the agent. The seeded version points at
ENVIRONMENT.md rather than repeating it, so the two can't drift.
Folding what it learned back into the seed
hermenix ctl pull ./hermes-home # excludes .env, sessions, state.db
diff -ru guest/hermes/home ./hermes-home
Worth doing occasionally. If the agent has written something genuinely good into its own SOUL.md, you want that in the seed before you ever rebuild from scratch.
Managing Hermes from the host
hermenix vm drives the virtual machine. hermenix ctl drives the agent
inside it.
hermenix ctl soul # current identity
hermenix ctl soul edit # pull, $EDITOR, push, offer to restart
hermenix ctl memory edit # memories/USER.md
hermenix ctl conf edit # config.yaml
hermenix ctl secrets # keys, values masked
hermenix ctl secrets edit # edit hermesHome's .env, written 0600
hermenix ctl secrets show # values, when you really mean it
hermenix ctl status # gateway service
hermenix ctl logs # follow it
hermenix ctl restart
hermenix ctl units # find the unit name if the default is wrong
hermenix ctl reseed # overwrite identity from the seed
hermenix ctl pull [dir] # copy live identity to the host
Anything unrecognised is passed straight through to the hermes CLI in the
guest, with a TTY attached — quote it as one string if it's more than a
single word (same rule as vm ssh above):
hermenix ctl chat
hermenix ctl update
hermenix ctl "sessions list"
hermenix ctl doctor
The edit variants pull the file out over vsock, open your $EDITOR
locally, write it back, and offer to restart the gateway. Secrets go in at
mode 0600 and never pass through the Nix store — which is the whole reason
there's no declarative option for them.
If status and logs come back empty, the gateway unit is probably named
something other than hermes-agent. Run hermenix ctl units and set
services.hermenix.gatewayUnit to match; upstream has used more than one
name for it.
Seeing what the guest is running
hermenix vm config
Prints the current system store path, the last ten generations, the git log
of /etc/nixos, and anything uncommitted. That last one matters: the agent
editing files without committing is the usual reason the config on disk and
the running system disagree.
To read or diff the config locally:
hermenix vm config-pull ./guest-live
diff -ru guest/ guest-live/
Expect that diff to grow. The guest/ directory in this repo is the seed
that was written on first boot; it is deliberately not kept in sync. Once the
agent edits its config, this repo and the guest diverge permanently, and a
host rebuild will not fix a guest config problem.
Letting the guest reach a service on your network
By default the guest can reach the public internet and nothing private. To open a single endpoint — say Ollama on another machine:
services.hermenix.lanAccess = [
{ address = "192.168.1.50"; port = 11434; } # ollama
{ address = "192.168.1.50"; port = 9000; proto = "udp"; }
];
Each entry punches one destination and port through the RFC1918 drop. The
guest reaches it as 192.168.1.50:11434 directly. Traffic is masqueraded, so
the other machine sees it arriving from your host's address, not the guest's.
There's also allowLan = true, which opens all private address space at
once. It exists, but reach for lanAccess instead — the whole point of this
design is that an agent which goes wrong can't go wandering.
A service on the host itself
Different chain, different option. ComfyUI on the host, port 8188:
services.hermenix.hostAccess = [ { port = 8188; } ];
The service must be bound somewhere the guest can see. If it listens
on 127.0.0.1 the rule won't help, because the guest arrives on the bridge
address. Bind it to 10.77.0.1 (bridge only, invisible to your LAN) or
0.0.0.0:
services.comfyui.listenAddress = "10.77.0.1"; # or whatever your service calls it
From the guest that's then http://10.77.0.1:8188.
Exposing a service the guest is running
Say the agent has built a web app on guest port 8080. Two halves.
Guest side. The agent opens its own firewall, in /etc/nixos:
networking.firewall.allowedTCPPorts = [ 22 8080 ];
Host side. Nothing at all, if you only want to reach it from the host — host-to-guest is outbound traffic and already permitted:
curl http://10.77.0.2:8080
To publish it to the wider network:
services.hermenix.publish = [
{ hostPort = 8080; guestPort = 8080; }
];
That DNATs the port on your uplink to the guest and opens the matching
forward rule. hostPort and guestPort don't have to match, so you can park
an awkward guest port somewhere sensible.
A note on this one: publishing a port is the one change here that lets the outside world start conversations with the agent's machine. Everything else in this design is the guest reaching out. Put a reverse proxy with TLS and auth in front of anything that isn't a toy, and keep the list short.
Putting the guest directly on your LAN
The default setup keeps the guest on an isolated bridge behind the host's
NAT, reachable from outside only through publish. This section is for
when you want the VM to be a first-class device on your LAN instead: its
own fixed address on the network, reachable on any port from any device,
while keeping everything it can see outbound locked down.
Enable it and tell it your LAN's numbers:
services.hermenix.network = {
bridged = true;
externalInterface = "enp191s0"; # your uplink (required)
# Static LAN addressing: bridged mode does not use your LAN's DHCP.
lanAddress = "10.0.12.50"; # the guest's fixed LAN address
hostLanAddress = "10.0.12.10"; # this host's own LAN address
lanGateway = "10.0.12.1"; # your router
lanPrefixLength = 24; # default 24
# lanDns = [ "1.1.1.1" ]; # optional; defaults to lanGateway
};
What that gives you: the bridge absorbs the uplink, the host and the guest
share the LAN, and each side has a fixed address. The guest never needs to
know its own address - the module runs a small DHCP server on the bridge
that reserves lanAddress for the guest's MAC, so the VM stays a plain
DHCP client and just receives its address. This host keeps its LAN identity
too: the bridge clones your uplink's MAC, so the LAN sees this host under
its usual MAC (no duplicate-address surprises), and the guest's DHCP
broadcast never leaves the bridge, so nothing on your LAN can race the
reservation.
What is and isn't reachable
- Inbound to the guest, from anywhere on the LAN: every port is open. The guest's own firewall decides what answers - by default it only opens port 22, so the agent opens whatever it wants LAN-reachable.
- Guest egress is still gated, exactly as in NAT mode:
lanAccessentries andallowLanare unchanged. The VM sees the LAN, but only the parts you have allowed. - Guest to host still works:
hostAccessand the host Nix cache are reachable over the service net at10.77.0.1in both modes. publishis ignored in this mode - every inbound port is already reachable. Drop the entries if you come back to NAT mode.
Switching over
-
Pick three free addresses: the guest (
lanAddress), this host (hostLanAddress- use the address it already has on that interface, e.g. a router reservation, so nothing else on the LAN breaks), and the router itself (lanGateway). KeephostLanAddressandlanAddressoutside any DHCP pool, or reserve them in the router first. -
Set
network.bridged = truewith the options above, thensudo nixos-rebuild switch. The network drops briefly while the host reconfigures - plan for it. -
Verify from the host:
hermenix vm ssh; inside the guest,ip -4 addrshows the service net (10.77.0.2) and its fixed LAN address. From any other device on the LAN:ssh hermes@<guest-lan-address> # and any other port the guest opens
Switching back
Set bridged = false (the lan* options are ignored) and rebuild the
host. The guest's DHCP client finds no server on the isolated bridge and
settles onto its static route via the host again; no guest-side change
needed.
Gotchas
hostLanAddressis a static claim on the LAN - pick one your router won't hand out, or reserve it.- Any network config you had for the uplink interface is replaced by its bridge-port role while bridged (this module's networkd unit for it takes priority over most others).
- An older guest image that predates the DHCP-on-NIC setup in the current
guest/common.nixwon't take an address in bridged mode; upgrade the image or apply that config inside the VM first.
Passing hardware to the guest
Before reaching for this
If what you want is "the agent can use the GPU", passthrough is often the wrong answer. It is exclusive — the host loses the card completely while the VM runs — and it is the single largest reduction in isolation available in this setup.
The alternative: keep Ollama or ComfyUI on the host where the GPU already
works, and open the port with hostAccess. The agent gets the capability
over HTTP and the device stays on your side of the boundary. That is usually
both easier and better.
Pass the hardware through when the guest genuinely needs the device itself — CUDA in-process, a capture card, a radio, a display output.
What passthrough costs you
A PCI device does DMA. The IOMMU is the only thing standing between a
passed-through device and host memory, which is why passthrough.iommu is a
hard requirement and the module refuses to build without it.
Two further constraints follow from how IOMMUs work:
-
Whole groups only. Every device in an IOMMU group goes together. Check yours before planning anything:
for d in /sys/kernel/iommu_groups/*/devices/*; do echo "group $(basename "$(dirname "$(dirname "$d")")"): $(lspci -nns "$(basename "$d")")" done | sort -VIf your GPU shares a group with a controller you need, you cannot pass it cleanly. ACS override patches exist; they work by lying about isolation, and they have no place in a setup built to contain an autonomous agent.
-
Every function. A GPU is at least two entries — graphics and its HDMI audio function.
GPU
lspci -Dnn | grep -i vga # addresses and vendor:device IDs
services.hermenix.passthrough = {
iommu = "amd"; # or "intel"
pci = [
{ address = "0000:01:00.0"; id = "10de:2204"; } # graphics
{ address = "0000:01:00.1"; id = "10de:1aef"; } # HDMI audio
];
};
Setting id makes vfio-pci claim the card in the initrd, before nvidia or
amdgpu can. Do this for GPUs — unbinding a live graphics driver afterwards
is unreliable, and the failure mode is a host with no display. Devices given
only an address are bound at boot by a oneshot service instead, which is
fine for things you swap around.
Rebuild and reboot — early binding happens in the initrd. Then check:
lspci -nnk -d 10de:2204 | grep "Kernel driver" # want: vfio-pci
On the guest side, the agent installs the drivers itself. It owns
/etc/nixos, so hardware.nvidia and nvidia-container-toolkit are its
problem, not yours — which is rather the point of the arrangement.
HDMI output
A display hanging off a passed-through GPU works once the guest drives the
card; nothing extra is needed here. A capture card or HDMI grabber is just
another PCI or USB device — find it with lspci -Dnn or lsusb and use the
matching section.
USB devices and Bluetooth
Matched by vendor and product ID:
lsusb # e.g. "Bus 001 Device 004: ID 8087:0026 Intel ..."
services.hermenix.passthrough.usb = [
{ vendorId = "8087"; productId = "0026"; } # bluetooth radio
{ vendorId = "0403"; productId = "6001"; } # serial adapter
];
This adds an xHCI controller to the guest and forwards the matching devices. No IOMMU requirement, no VFIO, and much milder than PCI passthrough — QEMU mediates rather than handing over raw hardware.
Two things that catch people:
- The host must not be using it. For an integrated Bluetooth radio, keep
btusboff it:boot.blacklistedKernelModules = [ "btusb" ];, and don't enablehardware.bluetoothon the host. - No hotplug. QEMU resolves these at start, so a device that's unplugged
when the VM boots won't appear when you plug it in.
hermenix vm restartafter plugging. Passing a whole USB controller through as a PCI device gets you real hotplug for anything in its ports, at the cost of that controller being exclusive to the guest.
When it doesn't start
hermenix vm console first. Then, in rough order of likelihood: the device
didn't bind to vfio-pci (lspci -nnk), the IOMMU isn't on (dmesg | grep -i iommu), or something in the group is still held by the host. The module
lifts LimitMEMLOCK for you when PCI passthrough is configured, since VFIO
pins the whole guest RAM allocation and the default limit otherwise produces
a memorably unhelpful error.
Virtual devices
The VM boots a fixed set of virtual hardware, and services.hermenix.devices
controls which optional pieces are attached. Everything here is a QEMU device:
none of it touches host hardware, and none of it needs anything the guest
doesn't already ship (the mainline virtio_* drivers).
| Option | Default | Attaches |
|---|---|---|
devices.rng.enable |
true |
virtio entropy source |
devices.vsock.enable |
true |
the host<->guest vsock (hermenix ctl, backups) |
devices.video.enable |
false |
a virtual display adapter |
devices.video.model |
"virtio" |
virtio (has a render node) or a legacy VGA model |
devices.balloon.enable |
false |
virtio memory balloon |
devices.watchdog.enable |
false |
hardware watchdog (i6300esb/ib700, action reset/…) |
devices.input.enable |
false |
virtio keyboard/tablet |
devices.audio.enable |
false |
audio device (intel-hda/virtio-sound) |
devices.tpm.enable |
false |
TPM 2.0 via swtpm |
devices.cdroms |
[] |
read-only CD images |
devices.extraDisks |
[] |
extra raw disks |
devices.usbController |
"auto" |
qemu-xhci (auto = on iff passthrough.usb is set) |
extraQemuArgs appends raw arguments to the QEMU command line for anything
not covered above. It is unvalidated - prefer a typed option where one exists.
A render node (Wayland, GPU compute, anything wanting /dev/dri)
services.hermenix.devices.video.enable = true; # model defaults to "virtio"
Gives the guest /dev/dri/card0 and /dev/dri/renderD128. The console stays
serial and nothing is shown on the host. Only the virtio model exposes the
render node; the legacy models give card0 alone.
TPM
devices.tpm.enable = true starts an swtpm unit first and points QEMU at
it. State lives in the unit's StateDirectory (/var/lib/hermes-vm-tpm by
default), so the guest gets a TPM 2.0 it can seal secrets to.
Audio, with a caveat
devices.audio.enable is wired up, but hermes-vm.service is deliberately
hardened (ProtectHome, PrivateTmp, a dedicated system user), so it cannot
reach your PipeWire socket without loosening that hardening. Leave it off
unless you mean to relax the unit for it.
Recovery
Four layers, cheapest first. The first two live inside the guest and cover almost every real mistake.
# The agent's own config history
hermenix ctl soul # what it thinks it is now
hermenix vm ssh "git -C /etc/nixos log --oneline"
# Previous NixOS generations, also in the guest's boot menu
hermenix vm ssh "sudo nixos-rebuild switch --rollback"
# Snapshot of the state disk
hermenix vm snapshot # trigger one now, e.g. before a risky change
hermenix vm snapshots
hermenix vm rollback state-20260914-1400 # asks you to type 'rollback' to confirm
# Restore from backup
hermenix vm backup # trigger one now
hermenix vm backups
Snapshots
Hourly by default, keeping 48. Only the state disk — the system disk isn't
snapshotted because hermenix setup rebuilds it from the flake in minutes.
The host fsfreezes the guest through qemu-guest-agent around each one, so the snapshot captures a consistent filesystem rather than the equivalent of pulling the power. Takes milliseconds; the VM keeps running.
hermenix vm rollback stops the VM, moves the current state subvolume aside as
state.old-<timestamp>, and restores the snapshot in its place. Nothing is
deleted, so a mistaken rollback is itself reversible. Clean the state.old-*
directories up yourself once you're happy.
Keep retention tighter than you would on ZFS. The state image is nodatacow, and every live snapshot reintroduces copy-on-write for the blocks it still references. Performance sags while snapshots exist and recovers as they're pruned.
Backups
services.hermenix.backup = {
enable = true;
repository = "sftp:backup@nas:/hermes";
passwordFile = "/var/lib/secrets/hermes-restic";
};
This backs up the guest's state, not its disk. Files are pulled over the SSH channel that already exists and streamed into restic.
Two reasons it works that way. The backup stays portable — it restores onto any filesystem, or a different machine entirely, and doesn't care that you're on btrfs today. And it avoids loop-mounting a guest-written filesystem on the host, which would mean the host kernel parsing data the agent controls. That would undo a fair bit of what the rest of this design is for.
If the repository is on your LAN rather than the internet, note this runs on
the host, so it needs no lanAccess entry — the guest never touches it.
When the guest won't boot at all
The system disk is disposable by design. Re-run hermenix setup, write a
fresh image to the system disk, and leave the state disk alone — Hermes comes
back with its identity, memories, workspace and NixOS config intact. That is
the intended repair, not a last resort.
Starting completely fresh
hermenix setup --clean stops the VM, deletes the existing
system.raw/state.raw/UEFI vars, and drops the stale known_hosts entry
for the guest (a fresh state disk means a fresh SSH host key on next boot,
so the old entry would otherwise fail with a host-key-mismatch on your next
connect) — then walks through setup again for a genuinely clean first boot,
rather than the disk-exists, skip-ahead path. It leaves the btrfs
subvolumes and any snapshots alone (only the files inside them), and asks
you to type clean to confirm before touching anything.
Nix flavour
The guest runs Determinate Nix, nixpkgs pinned to FlakeHub
(f/NixOS/nixpkgs/0.1, rolling; use 0 for current stable).
/etc/nix/nix.confbelongs to Determinate and must not be edited. Custom settings go in/etc/nix/nix.custom.conf, written fromguest/common.nix. There is deliberately nonix.settingsblock.- No
nix.gceither —determinate-nixdruns the collector. - Lazy trees are on by default, which makes the agent's flake evaluations much cheaper. That matters more than it sounds when something re-evaluates nixpkgs on a schedule.
- Neither flake uses
inputs.nixpkgs.followsfordeterminate. Determinate warn it causes FlakeHub Cache misses.
Substituter order in the guest: FlakeHub Cache, your host cache,
cache.nixos.org. If FlakeHub Cache goes quiet, the login has expired — run
determinate-nixd login again.
The host stays on stock Nix. Nothing here requires changing that.
What the guest can and cannot reach
| Default | |
|---|---|
| Public internet | Yes — it needs this for model APIs and messaging |
| Host Nix cache | Yes, one port, hostCache.enable = false closes it |
| Anything else on the host | No, unless in hostAccess |
| Your LAN and other VMs | No, unless in lanAccess |
| Inbound from the network | No, unless in publish |
| Host hardware | None, unless in passthrough |
With network.bridged = true the "Inbound from the network" row changes:
with the guest on the LAN, every port it opens is reachable from the LAN;
publish stops doing anything. The outbound rows are unchanged - the
lanAccess / allowLan gates still apply.
The guest owns its kernel and its config, so nothing inside it is a security
boundary. The nftables rules in modules/host.nix are. QEMU runs as an
unprivileged hermes-vm user with its own seccomp sandbox and systemd
hardening, with a thin device model: virtio-blk, virtio-net, vsock, RNG,
serial. No virtiofs. vhost-net is off deliberately — it would move packet
handling into the host kernel in exchange for throughput one agent doesn't
need.
The host cache does let the guest read every store path on the host. Store paths are world-readable anyway, so this isn't new exposure, but don't bake secrets into your host's store.
Known rough edges
- The Determinate NixOS module's option names for custom Nix settings aren't
published — only the nix-darwin
determinateNix.customSettingsequivalent is. Soguest/common.nixwrites/etc/nix/nix.custom.confviaenvironment.etc. If a future module version claims that file, you'll get a conflict; switch to its option then. determinate-nixd loginis manual. There's no way to bake the credential into the image without parking a token somewhere it shouldn't be.MemoryDenyWriteExecute = trueis correct under KVM but is the first thing to flip if QEMU refuses to start.pkgs.OVMF.firmware/pkgs.OVMF.variablesare current attribute names; older nixpkgs put these underpkgs.OVMF.fdin anFV/subdirectory.- PCI passthrough needs a reboot to take effect, because early vfio binding happens in the initrd.
- Snapshots need btrfs. Everything else works on any filesystem; without it you rely on backups alone for recovery.
state.old-*directories left byhermenix vm rollbackare not cleaned up automatically, on purpose. Delete them yourself once you're sure.- First boot builds a lot. Give it time.
services.hermenix.hermesHomecurrently defaults to a path that doesn't exist in the guest (it's built from the host's disk-image directory, not the guest's mounted state disk) — set it explicitly on every deployment, migrated or not. See "Migrating an existing deployment" below for the value to use.
Migrating an existing (e.g. Docker-based) Hermes deployment
If you already have a running Hermes with real data (HERMES_HOME pointing
at some directory full of SOUL.md, memories/, sessions/, state.db,
etc.):
nix run git+ssh://gitea@minipc1.lab.internal:22/ari/hermenix.git#hermenix -- migrate /path/to/existing/HERMES_HOME
Run ... -- migrate --help for the full set of options
(vsock CID, guest user, destination path, hostname rewrites). It copies —
never moves or deletes — the source data, so the original stays a safe
fallback until you've verified the new VM. To run it from a different
machine than the one hosting the VM, see hermenix remote above — the same
migrate /path/... argument goes in the quoted command string.
The module's services.hermenix.hermesHome option lets hermenix ctl (and
your own HERMES_HOME export inside the guest) point at wherever that
existing flat layout lands, instead of the default seeded .hermes under the
state disk.
Set hermesHome explicitly, even if you're not migrating. The option's
shipped default doesn't currently resolve to a real path inside the guest —
it's built from the host's own disk-image directory rather than the guest's
mounted state disk, so hermenix ctl soul/conf/secrets/etc. will fail
against it unhelpfully until you override it. For the default seeded layout:
services.hermenix.hermesHome = "/var/lib/hermes/.hermes";
For a migrated flat layout, use whatever --dest you passed to migrate
(default /var/lib/hermes).
Don't run it with sudo
migrate (like hermenix vm ssh and hermenix ctl) reaches the guest with
a plain ssh call over vsock, authenticating with whatever key or agent the
invoking user already has — the same identity that's in the guest's
authorizedKeys. Prefixing the command with sudo runs that ssh as
root instead, which has no matching key, and you'll hit:
hermes@hermes: Permission denied (publickey,keyboard-interactive).
If hermenix vm ssh works for you as your normal user, migrate doesn't
need anything more privileged — run it the same way.
If some of the source can't be read
Old flat HERMES_HOME directories, especially ones left behind by a
rootless-Docker deployment, sometimes contain a stray entry owned by a
container-internal uid with no mapping on the host — unreadable by anyone
outside the container. migrate tars the source with --ignore-failed-read,
so a warning like this is expected and non-fatal:
tar: ./.hermes: Cannot open: Permission denied
Everything else still gets copied and the run still completes, including the hostname rewrite step. If you actually need what's in the unreadable entry, grab it separately, as root, before or after the main migration — e.g.:
sudo tar -C /path/to/HERMES_HOME/.hermes -cf - . \
| hermenix vm ssh 'mkdir -p /var/lib/hermes/.hermes-old && tar -C /var/lib/hermes/.hermes-old -xf -'
hermes setup fails with "docker not found on PATH. Cannot route to container"
hermenix migrate automatically excludes .local/bin and .hermes-installed
from the migration tar, so this should not happen with current versions.
If you see it after a migration with an older build, the source's
.local/bin/hermes (a Docker-routing launcher from the previous environment)
was copied into the guest and is shadowing the native install.
Fix inside the guest:
# Remove the stale launcher
rm -f ~/.local/bin/hermes
# Reinstall the native hermes binary
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh \
| bash -s -- --non-interactive --skip-setup
# Verify it's a real binary, not a shell script
file ~/.local/bin/hermes
# Now proceed with setup
hermes setup
If hermes-install.service never ran because .hermes-installed was also
migrated from the source, you may instead let systemd do it:
rm -f ~/.local/bin/hermes ~/.hermes-installed
sudo systemctl start hermes-install
journalctl -fu hermes-install # watch until complete
hermes setup
Hacking on this repo
The devshell provides nushell, nufmt, and pre-commit hooks. Enter it once:
direnv allow
On first entry devenv installs the pre-commit hooks automatically. Hooks run on every commit, including:
- nufmt — formats all
*.nufiles in place - gen-completions — reruns
nu/gen-completions.nuand stagesnu/completions.nuwhenever a usage block in the source files changes - nu-tests — runs the CLI test suite (see below) whenever anything
under
nu/changes
Testing
Three test suites. None is ever run for you by an agent working in this repo — they're for you (or CI) to run.
CLI tests (nu/tests/), via nutest:
runtest # everything (bin/ is on PATH inside the devShell)
runtest vm_test # just nu/tests/vm_test.nu
runtest-debug vm_test # load vm_test.nu into an interactive nu session instead
Covers the CLI's pure helper functions (path/env defaults, shquote,
normalize-size, the gen-completions.nu parser) and a smoke test that the
whole hermenix/vm/ctl subcommand tree parses and dispatches correctly
— without ever invoking a subcommand's real body (no ssh, no disk writes,
no nix build).
Guest VM tests (guest/vm-tests/test_unbridge.nix / guest/vm-tests/test_bridge.nix):
boot guest/common.nix — the same module set hermes-guest-image is built
from — in ephemeral, throwaway NixOS test VMs. No real disk under
/var/lib/hermes-vm/** is ever touched:
runvmtest # NAT/isolated-bridge mode (plain pass/fail)
runvmtest guest-vm-bridged # bridged mode with a simulated LAN
runvmtest-debug # interactive driver REPL (ptpython) for the default suite
Both tests boot the guest the same way and share every check that must hold
in both modes — first-boot seeding (~/.hermes, ~/workspace, the
/etc/nixos symlink), that seeding is idempotent across a reboot (the whole
point of the .seeded marker), the hermes user/sudo setup, and SSH. That
shared node config and assertion script are defined once in
guest/vm-tests/common.nix and imported by both, so nothing can drift
between the two modes' common checks.
Each test then adds the checks that only make sense for its mode:
guest-vm-unbridged— the isolated bridge has no DHCP server, so it asserts the guest stands on its static config alone: exactly the one10.77.0.2/24address and exactly one default route, via the host.guest-vm-bridged— a second node simulates the LAN (a dnsmasq DHCP server, a throwaway HTTP service, and a plain device). It asserts the guest takes a DHCP lease beside the static service-net address, keeps both default routes (the LAN gateway at metric 100 and the host fallback at metric 200), can reach a service on the LAN, and that a LAN device can connect to a port the guest exposes (the guest's own firewall still decides what answers). The host-side half of bridged mode — the bridge itself, the mark-based rules — is covered statically by the check below.
Both deliberately do not cover the real hermes-install script (no internet
route inside the test VM) or modules/host.nix's own hermes-vm QEMU
service — those stay manual, operator-run checks.
runvmtest-debug gives you start_all(), machine.wait_for_unit(...), and
machine.shell_interact() (a live shell inside the guest) to step through a
failure by hand instead of just getting pass/fail.
Module evaluation checks (tests/network-modes-check.nix): a static
check — no build, no VM boot — that evaluates modules/host.nix in both
networking modes (NAT and network.bridged = true) plus guest/common.nix,
and asserts on the generated nftables rulesets, the firewall input/forward
rules, the networkd units, and the guest's DHCP/route-metric config:
nix build .#checks.x86_64-linux.network-modes
Catches regressions in the bridged-mode flag without booting anything. If it fails, the assertion that broke is named in the error.