add mars host, move Hermes Agent there from jupiter

New on-site host mars runs Hermes Agent as its sole service: joins the
tailnet, mounts jupiter's samba share at /mnt/jupiter (doubling as
Hermes's shared dropbox), and hosts state locally under /var/lib/hermes.
Same Authentik OIDC app/Telegram bot as before, just relocated — neptun's
hermes.mgaction.town vhost now points at mars.orbit.sol instead of jupiter.

hosts/jupiter/hermes-agent.nix and its three sops secrets are removed;
jupiter's Caddy vhost for it is gone too. Also refreshes tailscale_authkey
across all hosts and fixes two stale "erik@laptop" keys in flake.nix's
kexec/installer-iso images (leftover from a previous laptop, already
swapped out of common.nix back in 2fd5752) to darman@terra.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011FHr5ug9pu8q4XPrRkFnzJ

hosts/jupiter/configuration.nix

@@ -6,7 +6,6 @@
     ./hardware-configuration.nix
     ./disk-config.nix          # disko: OS-disk partitions + filesystems
     ./secrets.nix              # sops-nix: samba password, tailscale key, ...
-    ./hermes-agent.nix         # Hermes Agent, isolated instance (see file for why)
     ../../common.nix              # shared base: user / ssh / nix / firewall
     ../../services/network/samba.nix
     ../../services/network/avahi.nix

hosts/jupiter/hermes-agent.nix

@@ -1,83 +0,0 @@
-{ config, ... }:
-
-# Hermes Agent — jupiter's own instance (terra no longer runs one; see
-# 713d91d). Locked down given jupiter's much bigger blast radius
-# (irreplaceable immich photos on an unredundant RAID0, gitea/CI tokens, the
-# whole media stack):
-#
-#   - Own dedicated "hermes" system user (module default: user/group "hermes",
-#     createUser = true) — NOT darman. darman is in jupiter's "docker" group
-#     (services/containers.nix: rootful podman with dockerCompat), which is
-#     root-equivalent (`docker run -v /:/host --privileged ...`). Handing an
-#     LLM-driven agent that identity would mean a container escape = root on
-#     the whole NAS.
-#   - container.enable = true, backend = "podman": the ENTIRE gateway process
-#     runs inside a container (reusing jupiter's existing rootful podman
-#     instead of also standing up a second Docker daemon), not just the shell
-#     tool. Per upstream's own SECURITY.md this is "whole-process wrapping" —
-#     shell, file tools, MCP subprocesses, and the code-exec tool are all
-#     confined, unlike the lighter "terminal-backend"-only isolation.
-#   - Its own Telegram bot (own token, in secrets.nix) with an EXPLICIT
-#     TELEGRAM_ALLOWED_USERS rather than relying solely on the adapter's
-#     fail-closed default. Sharing terra's bot token would 409-conflict two
-#     long-pollers on the same token.
-#   - No container.extraVolumes into /mnt/data or the homelab repo — nothing
-#     valuable is in reach if a command goes wrong or gets injected via
-#     Telegram/tool output. stateDir/workingDirectory live on the array
-#     (below) purely because coding-task state (repo clones, npm/pip caches
-#     inside the container's writable layer) belongs off the 29G eMMC, same
-#     reasoning as postgres/containers.storage in configuration.nix — NOT
-#     because anything else on /mnt/data is exposed to the agent.
-{
-  services.hermes-agent = {
-    enable = true;
-    addToSystemPackages = true;   # `hermes` on darman's PATH for interactive
-                                   # debugging over ssh — routes through to the
-                                   # container, does not grant darman any group.
-
-    # Off the eMMC: stateDir bind-mounts into the container as /data, so this
-    # is where any future scoped repo clone (container.extraVolumes) and the
-    # container's own writable layer (npm/pip installs during coding tasks)
-    # actually land. RequiresMountsFor below (mirrors podman/sabnzbd/gitea-runner
-    # in configuration.nix) keeps the service from starting — and bind-mounting
-    # the wrong, empty eMMC path — before the nofail array is up.
-    stateDir = "/mnt/data/AppData/hermes";
-    workingDirectory = "/mnt/data/AppData/hermes/workspaces";
-
-    # HERMES_TIMEZONE is the highest-priority source hermes_time.py checks
-    # (ahead of config.yaml's `timezone` key) — the container has no host
-    # /etc/localtime bind-mount, so it defaults to UTC otherwise. Not a
-    # secret, so `environment` (plain .env) rather than sops.
-    environment.HERMES_TIMEZONE = "Europe/Berlin";
-
-    container = {
-      enable = true;
-      backend = "podman";   # jupiter already runs podman (services/containers.nix);
-                             # default "docker" would stand up a second daemon.
-    };
-
-    # OpenCode Go provider account (same key originally used for terra's now-
-    # removed instance, copied into secrets/jupiter.yaml — just an API key,
-    # not a stateful identity like the Telegram bot token).
-    settings.model = {
-      provider = "opencode-go";
-      base_url = "https://opencode.ai/zen/go/v1";
-      default = "gpt-5.6-luna";
-      api_mode = "codex_responses";
-    };
-
-    settings.platforms.telegram = {
-      enabled = true;
-      home_channel = {
-        platform = "telegram";
-        chat_id = "15151223";
-        name = "Erik Simon";
-        user_id = "15151223";
-      };
-    };
-
-    environmentFiles = [ config.sops.templates."hermes-agent.env".path ];
-  };
-
-  systemd.services.hermes-agent.unitConfig.RequiresMountsFor = [ "/mnt/data" ];
-}

hosts/jupiter/secrets.nix

@@ -64,21 +64,4 @@
   sops.secrets.sabnzbd_eweka_username.owner = "sabnzbd";
   sops.secrets.sabnzbd_eweka_password.owner = "sabnzbd";
 
-  # Hermes Agent (hosts/jupiter/hermes-agent.nix) — a separate, isolated
-  # instance from terra's, with its OWN Telegram bot token (sharing terra's
-  # would 409-conflict two long-pollers on one token). opencode_go_api_key
-  # is the same provider account as terra (hosts/terra/secrets.nix) — a
-  # stateless API key, fine to duplicate across hosts. No owner override:
-  # sops.templates renders via a root-run activation script, which the
-  # hermes module's own activation script (also root) then reads — unlike
-  # sabnzbd's preStart, this doesn't run as the service's own user.
-  sops.secrets.opencode_go_api_key = { };
-  sops.secrets.telegram_bot_token = { };
-  sops.templates."hermes-agent.env".content = ''
-    OPENCODE_GO_API_KEY=${config.sops.placeholder.opencode_go_api_key}
-    TELEGRAM_BOT_TOKEN=${config.sops.placeholder.telegram_bot_token}
-    TELEGRAM_HOME_CHANNEL=15151223
-    TELEGRAM_ALLOWED_USERS=15151223
-  '';
-
 }

hosts/mars/configuration.nix

@@ -0,0 +1,49 @@
+{ config, pkgs, ... }:
+
+# mars — on-site x86_64 box, single-purpose: runs Hermes Agent only.
+# See hermes-agent.nix for what that is and why it moved here from jupiter.
+{
+  imports = [
+    ./hardware-configuration.nix
+    ./disk-config.nix          # disko: OS-disk partitions + filesystems
+    ./secrets.nix              # sops-nix: samba/tailscale/hermes secrets
+    ./hermes-agent.nix
+    ../../common.nix              # shared base: user / ssh / nix / firewall
+    ../../services/containers.nix
+    ../../services/vpn/tailscale.nix
+  ];
+
+  networking.hostName = "mars";
+  networking.networkmanager.enable = true;   # DHCP on-site, same as jupiter
+  users.users.darman.extraGroups = [ "docker" ];   # merges with common.nix; podman debug access
+
+  # ---- Boot (UEFI, confirmed) ----
+  boot.loader.systemd-boot.enable = true;
+  boot.loader.efi.canTouchEfiVariables = true;
+
+  # jupiter's samba share (services/network/samba.nix) — mounted on demand so
+  # mars doesn't stall boot/login when jupiter is off or unreachable. This is
+  # also where Hermes's shared dropbox lives now (hermes-agent.nix). Modes are
+  # tighter than terra's equivalent mount (0770 not 0755, gid=hermes not
+  # gid=users) since the hermes-agent container (uid 986, gid 983 — no podman
+  # userns remapping, see services/network/pihole.nix) needs group write into
+  # it, not just darman.
+  fileSystems."/mnt/jupiter" = {
+    device = "//jupiter/data";
+    fsType = "cifs";
+    options = [
+      "credentials=${config.sops.templates."jupiter-smb.credentials".path}"
+      "uid=1000"
+      "gid=983"
+      "file_mode=0770"
+      "dir_mode=0770"
+      "nofail"
+      "x-systemd.automount"
+      "x-systemd.idle-timeout=60"
+      "x-systemd.mount-timeout=10s"
+      "_netdev"
+    ];
+  };
+
+  system.stateVersion = "26.05";   # set at install time; do NOT bump on upgrades
+}

hosts/mars/disk-config.nix

@@ -0,0 +1,37 @@
+{ ... }:
+
+# Declarative OS-disk layout (disko). UEFI: GPT with an ESP + ext4 root,
+# same pattern as jupiter/terra (confirmed UEFI-capable, not the legacy-BIOS
+# guess this scaffold started with).
+#
+# ⚠️  This disk is WIPED on install. Set `device` below to the real OS disk
+#     ONLY (by-id) — `ls -l /dev/disk/by-id` once you have console access.
+{
+  disko.devices.disk.os = {
+    type = "disk";
+    device = "/dev/disk/by-id/ata-Samsung_SSD_840_EVO_120GB_S1D5NSAFB10834Z";
+    content = {
+      type = "gpt";
+      partitions = {
+        ESP = {
+          size = "512M";
+          type = "EF00";
+          content = {
+            type = "filesystem";
+            format = "vfat";
+            mountpoint = "/boot";
+            mountOptions = [ "umask=0077" ];
+          };
+        };
+        root = {
+          size = "100%";
+          content = {
+            type = "filesystem";
+            format = "ext4";
+            mountpoint = "/";
+          };
+        };
+      };
+    };
+  };
+}

hosts/mars/hardware-configuration.nix

@@ -0,0 +1,18 @@
+# Do not modify this file!  It was generated by ‘nixos-generate-config’
+# and may be overwritten by future invocations.  Please make changes
+# to /etc/nixos/configuration.nix instead.
+{ config, lib, pkgs, modulesPath, ... }:
+
+{
+  imports =
+    [ (modulesPath + "/installer/scan/not-detected.nix")
+    ];
+
+  boot.initrd.availableKernelModules = [ "xhci_pci" "ehci_pci" "ahci" "usbhid" "usb_storage" "sd_mod" ];
+  boot.initrd.kernelModules = [ ];
+  boot.kernelModules = [ "kvm-intel" ];
+  boot.extraModulePackages = [ ];
+
+  nixpkgs.hostPlatform = lib.mkDefault "x86_64-linux";
+  hardware.cpu.intel.updateMicrocode = lib.mkDefault config.hardware.enableRedistributableFirmware;
+}

hosts/mars/hermes-agent.nix

@@ -0,0 +1,146 @@
+{ config, ... }:
+
+# Hermes Agent — moved here from jupiter (hosts/jupiter/hermes-agent.nix,
+# see its git history / b5fa599 / 713d91d for the terra->jupiter->mars
+# lineage). mars is dedicated to this one service, on-site, with no big
+# data array of its own — unlike jupiter it has nothing under /mnt/data, so
+# state lives on the local OS disk and the shared dropbox rides jupiter's
+# samba share as a CIFS client instead of being served locally.
+#
+# Runs the OFFICIAL published image (docker.io/nousresearch/hermes-agent —
+# real and actively maintained, contrary to what the checked-out repo's own
+# README/docker-compose.yml suggested; verified directly on Docker Hub) as a
+# plain podman container. It never sets HERMES_MANAGED or writes .managed, so
+# Hermes fully self-manages config.yaml, profiles, memories and skills at
+# runtime — no redeploy needed except to bump the pinned digest below.
+#
+# Security posture:
+#   - Only two paths reachable: its own local state dir, and the small
+#     shared "dropbox" below (via the jupiter samba mount) for darman to
+#     hand files to Hermes — nothing else on jupiter's array is reachable
+#     if a command goes wrong or gets injected via Telegram/tool output.
+#   - Its own Telegram bot (own token, in secrets.nix) with an EXPLICIT
+#     TELEGRAM_ALLOWED_USERS.
+#   - Runs as a rootful podman container (services/containers.nix) with its
+#     OWN numeric uid/gid — not darman, who is in the "hermes" group for
+#     host-level debugging only (`podman exec -it hermes-agent hermes ...`).
+#
+# Dashboard (HERMES_DASHBOARD=1) is gated behind Authentik, same setup as on
+# jupiter. Its default bind (0.0.0.0:9119) fails closed without an auth
+# provider registered, and 0.0.0.0 (not loopback) is required so neptun's
+# Caddy can reach it over tailscale0 — reachability itself stays LAN-closed
+# (no networking.firewall.allowedTCPPorts entry; tailscale0 is already a
+# trustedInterface, services/vpn/tailscale.nix). Public route: neptun's
+# hermes.mgaction.town vhost (hosts/neptun/configuration.nix) proxies to this
+# over the tailnet. mars runs no Caddy of its own (single-purpose box), so
+# there is no LAN vhost — reach the dashboard directly via mars's tailnet
+# name (mars.orbit.sol:9119) or LAN IP:9119 for local debugging.
+#
+# Uses upstream's generic self-hosted OIDC plugin, same Authentik
+# application as before (slug `hermes`) — the client ID/secret didn't need
+# to change since the public redirect URI (hermes.mgaction.town) didn't.
+#
+# Data migration: this starts with a FRESH state dir. jupiter's instance was
+# itself reset to fresh on 2026-08-21 (see its old hermes-agent.nix), so
+# there was nothing irreplaceable to carry forward; if that turns out to be
+# wrong, jupiter's old data is backed up at
+# /mnt/data/AppData/hermes.bak-2026-08-21 and can be rsynced into
+# ${hermesHome} below before the first switch on mars.
+let
+  stateDir = "/var/lib/hermes";
+  hermesHome = "${stateDir}/.hermes";
+  # Shared drop-in folder: darman can put files here from any host. Lives on
+  # jupiter's array (reachable at /mnt/jupiter, the samba mount below) rather
+  # than locally, so it's the same physical location it always was — only
+  # the container reading it moved. Mounted under /opt/data so it falls
+  # inside Hermes's own sealed write-safe root (HERMES_WRITE_SAFE_ROOT=
+  # /opt/data) rather than a path its own tooling would treat as untrusted.
+  dropboxDir = "/mnt/jupiter/AppData/hermes-dropbox";
+
+  # Pinned by digest (captured 2026-08-21 via `podman image inspect
+  # docker.io/nousresearch/hermes-agent:latest --format '{{.Digest}}'` on
+  # jupiter) rather than floating `:latest`, so a redeploy is reproducible —
+  # bumping Hermes is an explicit edit here, not silent drift on next pull.
+  hermesImage = "docker.io/nousresearch/hermes-agent@sha256:5342e518734a08f6c66b89b4262434813c28a77abbc59c230c8f1637df71a259";
+
+  # Kept identical to jupiter's instance purely so nothing else needs to
+  # change if state ever gets migrated over.
+  hermesUid = "986";
+  hermesGid = "983";
+in
+{
+  # Browsing convenience (ssh access to the bind-mounted local state) — does
+  # NOT touch the container, which keeps using HERMES_UID/GID above
+  # regardless of what's declared here.
+  users.groups.hermes.gid = 983;
+  users.users.darman.extraGroups = [ "hermes" ];
+
+  systemd.tmpfiles.rules = [
+    "d ${stateDir} 0750 root hermes -"
+  ];
+
+  # podman requires the bind-mount source to already exist (no auto-create),
+  # and the dropbox lives on the CIFS mount below — mkdir there works fine
+  # over cifs, no server-side (jupiter) config needed.
+  systemd.services.hermes-agent-prepare-dirs = {
+    description = "Create Hermes state dirs before the container starts";
+    before = [ "podman-hermes-agent.service" ];
+    wantedBy = [ "podman-hermes-agent.service" ];
+    unitConfig.RequiresMountsFor = [ "/mnt/jupiter" ];
+    serviceConfig.Type = "oneshot";
+    script = ''
+      mkdir -p ${hermesHome}
+      mkdir -p ${dropboxDir}
+    '';
+  };
+
+  virtualisation.oci-containers.containers.hermes-agent = {
+    image = hermesImage;
+    autoStart = true;
+    # Host networking: Hermes only long-polls Telegram outbound, no inbound
+    # ports to publish (same reasoning as clonarr on jupiter).
+    extraOptions = [ "--network=host" ];
+    # Upstream's own documented single-mount pattern (docker/docker-compose.yml):
+    # ~/.hermes:/opt/data.
+    volumes = [
+      "${hermesHome}:/opt/data"
+      "${dropboxDir}:/opt/data/dropbox"
+    ];
+    environment = {
+      HERMES_UID = hermesUid;
+      HERMES_GID = hermesGid;
+      TZ = "Europe/Berlin";
+      # HERMES_TIMEZONE is the highest-priority source hermes_time.py checks
+      # (ahead of config.yaml's `timezone` key) — the container has no host
+      # /etc/localtime bind-mount, so it defaults to UTC otherwise (fixed in
+      # 9403122 on jupiter; carried forward here).
+      HERMES_TIMEZONE = "Europe/Berlin";
+
+      # Dashboard + Authentik OIDC gate — see the file-level comment above.
+      HERMES_DASHBOARD = "1";
+      HERMES_DASHBOARD_HOST = "0.0.0.0"; # must be tailscale0-reachable, not just loopback
+      HERMES_DASHBOARD_OIDC_ISSUER = "https://auth.mgaction.town/application/o/hermes/";
+      HERMES_DASHBOARD_OIDC_CLIENT_ID = "4BqdJu3htnMtSZnyEu5zHnsSOvlEbw3Ie3mYVlh6";
+      # uvicorn's proxy_headers=True (web_server.py) only trusts
+      # X-Forwarded-Proto from forwarded_allow_ips, which defaults to
+      # 127.0.0.1 — neptun's Caddy reaches this over the tailnet (a real
+      # routed IP), so without this the dashboard sees the raw scheme (http)
+      # and builds an http:// redirect_uri that Authentik rejects against its
+      # registered https:// one. Safe to trust any peer here: 9119 is already
+      # scoped to loopback + tailscale0 only (no LAN firewall rule), so
+      # nothing untrusted can reach this process to begin with.
+      FORWARDED_ALLOW_IPS = "*";
+    };
+    environmentFiles = [ config.sops.templates."hermes-agent.env".path ];
+    cmd = [ "gateway" "run" ];
+  };
+
+  systemd.services.podman-hermes-agent = {
+    after = [
+      "hermes-agent-prepare-dirs.service"
+      "systemd-tmpfiles-setup.service"
+    ];
+    requires = [ "hermes-agent-prepare-dirs.service" ];
+    unitConfig.RequiresMountsFor = [ "/mnt/jupiter" ];
+  };
+}

hosts/mars/secrets.nix

@@ -0,0 +1,38 @@
+{ config, ... }:
+
+# sops-nix wiring for mars. Encrypted values in ../../secrets/mars.yaml,
+# decrypted with mars's own SSH host key (recipient in ../../.sops.yaml).
+# The host key is pre-generated on the laptop and shipped at install
+# (nixos-anywhere --extra-files -> /etc/ssh/ssh_host_ed25519_key).
+{
+  sops.defaultSopsFile = ../../secrets/mars.yaml;
+  sops.age.sshKeyPaths = [ "/etc/ssh/ssh_host_ed25519_key" ];
+
+  sops.secrets.darman_password.neededForUsers = true;
+  users.users.darman.hashedPasswordFile = config.sops.secrets.darman_password.path;
+
+  sops.secrets.tailscale_authkey = { };
+
+  # Credentials file for the //jupiter/data cifs mount (see configuration.nix).
+  # Same value as jupiter's own samba_password (services/network/samba.nix) —
+  # mars authenticates as the same smb user, mirroring terra's setup.
+  sops.secrets.samba_password = { };
+  sops.templates."jupiter-smb.credentials".content = ''
+    username=darman
+    password=${config.sops.placeholder.samba_password}
+  '';
+
+  # Hermes Agent (hermes-agent.nix) — moved here from jupiter (see that
+  # host's git history); same Telegram bot token, opencode key, and
+  # Authentik OIDC client secret, so no new bot/app to provision.
+  sops.secrets.opencode_go_api_key = { };
+  sops.secrets.telegram_bot_token = { };
+  sops.secrets.hermes_dashboard_oidc_client_secret = { };
+  sops.templates."hermes-agent.env".content = ''
+    OPENCODE_GO_API_KEY=${config.sops.placeholder.opencode_go_api_key}
+    TELEGRAM_BOT_TOKEN=${config.sops.placeholder.telegram_bot_token}
+    TELEGRAM_HOME_CHANNEL=15151223
+    TELEGRAM_ALLOWED_USERS=15151223
+    HERMES_DASHBOARD_OIDC_CLIENT_SECRET=${config.sops.placeholder.hermes_dashboard_oidc_client_secret}
+  '';
+}

hosts/neptun/configuration.nix

@@ -110,6 +110,13 @@
     reverse_proxy http://jupiter.orbit.sol:2283
   '';
 
+  # ---- Hermes dashboard ----
+  # Authentik-gated (hosts/mars/hermes-agent.nix has the OIDC config and the
+  # "create the Authentik app" instructions — moved here from jupiter).
+  services.caddy.virtualHosts."hermes.mgaction.town".extraConfig = ''
+    reverse_proxy http://mars.orbit.sol:9119
+  '';
+
   # ---- Gitea WebUI ----
   # Gitea's web UI and HTTPS clones (services/dev/gitea.nix, HTTP_PORT 3000).
   # Its SSH side is the separate :2222 forward further down.