Wag adds MFA, route restriction and device enrolment to wireguard.

Key Features:

• Define routes which require MFA authorisation, or public always accessible routes
• Easy API for registering new clients
• High Availability
• Multiple MFA options, including webauthn, oidc and more

This work was very kindly supported by Aura Information Security.

iptables and libpam must be installed.
Wag must be run as root, to manage iptables and the wireguard device.

Forwarding must be enabled in sysctl.

sysctl -w net.ipv4.ip_forward=1

Wag does not need wg-quick or other equalivent as long as the kernel supports wireguard.

Both options require a kernel newer than 5.9+

Binary release (requires glibc 2.31+):

curl -L $(curl -s https://api.github.com/repos/NHAS/wag/releases/latest | jq -M -r '.assets[0].browser_download_url') -o wag
sudo ./wag gen-config

sudo ./wag start -config <generated_config_name>

From source (will require go1.19, npm, gulp, clang, llvm-strip, libbpf):

git clone git@github.com:NHAS/wag.git
cd wag
make

cp example_config.json config.json

sudo ./wag start

If running behind a reverse proxy, X-Forwarded-For must be set.

The root user is able to manage the wag server with the following command:

wag subcommand [-options]

Supported commands: start, cleanup, reload, version, firewall, registration, devices, users, webadmin, gen-config

start: starts the wag server

Usage of start:
  Start wag server (does not daemonise)
  -join string
        Cluster join token
  -config string
        Configuration file location (default "./config.json")

cleanup: Will remove all firewall forwards, and shutdown the wireguard device

reload: Reloads ACLs from configuration

version: Display the version of wag

firewall: Get firewall rules

Usage of firewall:
  -list
        List firewall rules
  -socket string
        Wag socket to act on (default "/tmp/wag.sock")

registration: Deals with creating, deleting and listing the registration tokens

Usage of registration:
  -add
        Create a new enrolment token
  -del
        Delete existing enrolment token
  -group value
        Manually set user group (can supply multiple -group, or use -groups for , delimited group list, useful for OIDC)
  -groups string
        Set user groups manually, ',' delimited list of groups, useful for OIDC
  -list
        List tokens
  -overwrite string
        Add registration token for an existing user device, will overwrite wireguard public key (but not 2FA)
  -socket string
        Wag socket to act on (default "/tmp/wag.sock")
  -token string
        Manually set registration token (Optional)
  -username string
        User to add device to

devices: Manages devices

Usage of devices:
  -address string
        Address of device
  -del
        Remove device and block wireguard access
  -list
        List wireguard devices
  -lock
        Lock device access to mfa routes
  -mfa_sessions
        Get list of devices with active authorised sessions
  -socket string
        Wag control socket to act on (default "/tmp/wag.sock")
  -unlock
        Unlock device
  -username string
        Owner of device (indicates that command acts on all devices owned by user)

users: Manages users MFA and can delete all users devices

Usage of users:
  -del
        Delete user and all associated devices
  -list
        List users, if '-username' supply will filter by user
  -lockaccount
        Lock account disable authention from any device, deauthenticates user active sessions
  -reset-mfa
        Reset MFA details, invalids all session and set MFA to be shown
  -socket string
        Wag socket location, (default "/tmp/wag.sock")
  -unlockaccount
        Unlock a locked account, does not unlock specific device locks (use device -unlock -username <> for that)
  -username string
        Username to act upon

webadmin: Manages the administrative users for the web UI

Usage of webadmin:
  -add
        Add web administrator user (requires -password)
  -del
        Delete admin user
  -list
        List web administration users, if '-username' supply will filter by user
  -lockaccount
        Lock admin account disable login for this web administrator user
  -password string
        Username to act upon
  -socket string
        Wag instance control socket (default "/tmp/wag.sock")
  -unlockaccount
        Unlock a web administrator account
  -username string
        Admin Username to act upon

1. Copy wag, config.json to /opt/wag
2. Generate a wireguard private key with wg genkey set PrivateKey in the example config to it
3. Copy (or link) wag.service to /etc/systemd/system/ and start/enable the service

First generate a token.

# ./wag registration -add -username tester
token,username
e83253fd9962c68f73aa5088604f3f425d58a963bfb5c0889cca54d63a34b2e3,tester

Then curl said token.

curl http://public.server.address:8080/register_device?key=e83253fd9962c68f73aa5088604f3f425d58a963bfb5c0889cca54d63a34b2e3

The service will return a fully templated response:

[Interface]
PrivateKey = <omitted>
Address = 192.168.1.1

[Peer]
Endpoint =  public.server.address:51820
PublicKey = pnvl40WiRt++0NucEGexlpfwWA8QzBYg2+8ZWZJvejA=
AllowedIPs = 10.7.7.7/32, 192.168.1.1/32, 192.168.3.4/32, 192.168.3.5/32
PersistentKeepAlive = 10

Which can then be written to a config file.

To authenticate the user should browse to the servers vpn address, in the example, case 192.168.1.1:8080, where they will be prompted for their 2fa code.
The configuration file specifies how long a session can live for, before expiring.

Make sure that you have ManagementUI.Enabled set as true, then do the following from the console:

sudo ./wag webadmin -add -username <your_username> -password <your-password-here>

Then browse to your management listening address and enter your credentials.

The web interface itself cannot add administrative users.

NumberProxies: The number of trusted reverse proxies before the client, makes wag respect the X-Forward-For directive and parses the client IP from it correctly HelpMail: The email address that is shown on the prompt page
Lockout: Number of times a person can attempt mfa authentication before their account locks
NAT: Turn on or off masquerading
ExposePorts: Expose ports on the VPN server to the client (adds rules to IPtables) example: [ "443/tcp", "100-200/udp" ]
CheckUpdates: If enabled (off by default) the management UI will show an alert if a new version of wag is available. This talks to api.github.com
MFATemplatesDirectory: A string path option, when set templates will be queried from disk rather than the embedded copies. Allows you to customise the MFA registration, entry, and success pages, allows custom js and css in the MFATemplatesDirectory /custom/ directory
DownloadConfigFileName: The filename of the wireguard config that is downloaded, defaults to wg0.conf

ExternalAddress: The public address of the server, the place where wireguard is listening to the internet, and where clients can reach the /register_device endpoint

MaxSessionLifetimeMinutes: After authenticating, a device will be allowed to talk to privileged routes for this many minutes, if -1, timeout is disabled
SessionInactivityTimeoutMinutes: If a device has not sent data in n minutes, it will be required to reauthenticate, if -1 timeout is disabled

DatabaseLocation: Where to load the sqlite3 database from, it will be created if it does not exist
Socket: Wag control socket, changing this will allow multiple wag instances to run on the same machine
Acls: Defines the Groups and Policies that restrict routes
Policies: A map of group or user names to policy objects which contain the wag firewall & route capture rules. The most specific match governs the type of access a user has to a route, e.g if you have a /16 defined as MFA, but one ip address in that range as allow that is /32 then the /32 will take precedence over the /16
Policies.<policy name>.Mfa: The routes and services that require Mfa to access
Policies.<policy name>.Public: Routes and services that do not require authorisation Policies.<policy name>.Deny: Deny access to this route

Webserver: Object that contains the public and tunnel listening addresses of the webserver

WebServer.Public.ListenAddress: Listen address for endpoint
WebServer.Tunnel.Port: Port for in-vpn-tunnel webserver, this does not take a full IP address, as the tunnel listener should never be outside the wireguard device

WebServer.<endpoint>.CertPath: TLS Certificate path for endpoint
WebServer.<endpoint>.KeyPath: TLS key for endpoint

Authenticators: Object that contains configurations for the authentication methods wag provides
Authenticators.Issuer: TOTP issuer, the name that will get added to the TOTP app
Authenticators.DomainURL: Full url of the vpn authentication endpoint, required for webauthn and oidc Authenticators.DefaultMethod: String, default method the user will be presented, if not specified a list of methods is displayed to the user (possible values: webauth, totp, oidc, pam)
Authenticators.Methods: String array, enabled authentication methods, e.g ["totp","webauthn","oidc", "pam"].

Authenticators.OIDC: Object that contains OIDC specific configuration options Authenticators.OIDC.IssuerURL: Identity provider endpoint, e.g http://localhost:8080/realms/account Authenticators.OIDC.ClientID: OIDC identifier for application Authenticators.OIDC.ClientSecret: OIDC secret Authenticators.OIDC.GroupsClaimName: Not yet used.

Authenticators.PAM.ServiceName: Name of PAM-Auth file in /etc/pam.d/ will default to /etc/pam.d/login if unset or empty

Clustering: Object containing the clustering details
Clustering.ClusterState: Same as the etcd cluster state setting, can be either new, create a new cluster, or existing. If you are joining an existing cluster, use start -join rather than this
Clustering.ETCDLogLevel: Level of logging for the embedded etcd server to emit, options info, error
Clustering.Witness: Is the node a witness node, i.e one that does not start a wireguard device, or management UI, but replicates events for the RAFT concensus
Clustering.TLSManagerListenURL: URL for generating certificates for the wag cluster, must be reachable by all nodes, typically automatically set by start -join

Wireguard: Object that contains the wireguard device configuration
Wireguard.DevName: The wireguard device to attach or to create if it does not exist, will automatically add peers (no need to configure peers with wg-quick)
Wireguard.ListenPort: Port that wireguard will listen on
Wireguard.PrivateKey: The wireguard private key, can be generated with wg genkey
Wireguard.Address: Subnet the VPN is responsible for
Wireguard.MTU: Maximum transmissible unit defaults to 1420 if not set for IPv4 over Ethernet
Wireguard.DNS: An array of DNS servers that will be automatically used, and set as "Allowed" (no MFA)

ManagementUI: Object that contains configurations for the webadministration portal. It is not recommend to expose this portal, I recommend setting ListenAddress to 127.0.0.1/localhost and then use ssh forwarding to expose it
ManagementUI.Enabled: Enable the web UI
ManagementUI.ListenAddress: Listen address to expose the management UI on
ManagementUI.CertPath: TLS Certificate path for management endpoint
ManagementUI.KeyPath: TLS key for the management endpoint

Full config example

{
    "Proxied": true,
    "ExposePorts": [
        "443/tcp",
        "100-200/udp"
     ],
    "CheckUpdates": true,
    "Lockout": 5,
    "NAT": true,
    "HelpMail": "help@example.com",
    "MaxSessionLifetimeMinutes": 2,
    "SessionInactivityTimeoutMinutes": 1,
    "ExternalAddress": "81.80.79.78",
    "DatabaseLocation": "devices.db",
    "Socket":"/tmp/wag.sock",
    "Webserver": {
        "Public": {
            "ListenAddress": "192.168.121.61:8080",
            "CertPath": "/etc/example/cert/path",
            "KeyPath": "/etc/ssl/private/somecert.key"
        },
        "Tunnel": {
            "Port": "8080"
        }
    },
    "ManagementUI": {
        "ListenAddress": "127.0.0.1:4433",
        "CertPath": "/etc/example/cert/path",
        "KeyPath": "/etc/ssl/private/somecert.key",
        "Enabled": true
    },
    "Authenticators": {
        "Issuer": "vpn.test",
        "DomainURL": "https://vpn.test:8080",
        "DefaultMethod":"webauthn",
        "Methods":["totp","webauthn", "oidc", "pam"],
        "OIDC": {
            "IssuerURL": "http://localhost:8080/",
            "ClientSecret": "<OMITTED>",
            "ClientID": "account",
            "GroupsClaimName": "groups"
        }
    },
    "Clustering": {
        "ClusterState": "new",
        "ETCDLogLevel": "error",
        "Witness": false,
        "TLSManagerListenURL": "https://wag.server:3434"
    },
    "Wireguard": {
        "DevName": "wg0",
        "ListenPort": 53230,
        "PrivateKey": "AN EXAMPLE KEY",
        "Address": "192.168.1.1/24",
        "MTU": 1420,
        "DNS": ["1.1.1.1"]
    },
    "Acls": {
        "Groups": {
            "group:nerds": [
                "daviv.test",
                "franky.someone",
                "any_username"
            ]
        },
        "Policies": {
            "*": {
                "Mfa": [
                     "10.0.0.2/32 8080/any"
                ],
                "Allow": [
                    "10.7.7.7/32",
                    "google.com"
                ]
            },
            "username": { 
                "Mfa": [
                     "someinternal.service 9100/tcp"
                ],
                "Allow":[ "10.0.0.1/32"]
            },
            "group:nerds": {
                "Mfa": [
                    "192.168.3.4/32",
                    "10.0.0.0/24",
                    "thing.internal 443/tcp icmp"
                ],
                "Allow": [
                    "192.168.3.5/32"
                ],
                "Deny": [
                    "10.0.0.5/32"
                 ]
            }
        }
    }
}

The Policies section allows you to define what routes should be both captured by the VPN and what ports and protocols are allowed through Wag.

Rules use the subnet prefix length to determine which rule applies. The most specific match is use to determine the level of user access to a route.
For example:

 "*": {
                "Mfa": [
                     "10.0.0.0/16"
                ],
                "Allow": [
                    "10.0.1.1/32",
                ]
            },

Users will be able to access 10.0.1.1 without MFA as the match is more specific. This change occured in v6.0.0, previously MFA routes would always take precedence.

Additionally if multiple policies are defined for a single route they are composed with MFA rules taking preference.
For example:

 "*": {
            "Mfa": [
                  "10.0.0.0/16",
                  "10.0.1.1/32 22/tcp",
            ]
  },
 "group:users": {
            "Allow": [
                  "10.0.1.1/32 443/tcp",
            ]
 }

All users will be able to access 22/tcp on the 10.0.1.1/32 host, but users in the group:users will be able to access 443/tcp on that host as well, along with 22/tcp when authorized.

As of [version number, yet to be released] you can now define deny rules which will block access to a route.

Example:

 "*": {
            "Allow": [
                  "10.0.0.0/16",
                  "10.0.1.1/32 443/tcp",
            ]
  },
 "group:users": {
            "Deny": [
                  "10.0.1.1/32 443/tcp",
            ]
 }

Its important to note that the most specific rule effectively creates a new rule "bucket", so if you do something like:

"group:nerds": {
      "Allow": [
            "10.0.0.0/24 443/tcp"
      ],
      "Deny": [
            "10.0.0.5/32 22/tcp"
      ]
}

Your clients will not be able to access 10.0.0.5/32 443/tcp, as the only rule in the /32 "bucket" is a deny rule. You can solve this by adding the following:

"group:nerds": {
      "Allow": [
            "10.0.0.0/24 443/tcp"
            "10.0.0.5/32 22/tcp"
      ],
      "Deny": [
            "10.0.0.5/32 22/tcp"
      ]
}

or

"group:nerds": {
      "Allow": [
            "10.0.0.0/24 443/tcp"
      ],
      "Deny": [
            "10.0.0.0/24 22/tcp"
      ]
}

As then you're adding the deny rule to the /24 "bucket".

Additionally, It is possible to define what services a user can access by defining port and protocol rules.
Currently 3 types of port and protocol rules are supported:

When no other rules are defined or the any keyword is used wag will allow all services and port combinations.

Example:

"1.1.1.1": Allows all ports and protocols to 1.1.1.1/32
"1.1.1.1 54/any": Allows both tcp and udp to 1.1.1.1/32

Example:

192.168.1.1 22/tcp 53/udp: Fairly self explanatory, allows you to hit 22/tcp and 53/udp on a host
1.1.1.1 icmp: As icmp doesnt have ports really you dont need it either

You can also define a range of ports with a protocol. wag requires that the lower port is first.

Example:

192.168.1.1 22-1024/tcp 23-53/any: Format is low port-high port/service

• Only supports clients with one AllowedIP, which is perfect for site to site, or client -> server based architecture.
• IPv4 only.
• Linux only
• Very Modern kernel 5.9+ at least (>5.9 allows loops in ebpf and bpf_link)

With the introduction of the MFATemplatesDirectory option, you can now specify a directory that contains template files for customising the MFA entry, registration and wireguard config file.
An example of all these files can be found in the embedded variants here: internal/webserver/resources/templates.

When the option is set, you must define all the files this guide is a brief description of what each file is:
interface.tmpl: The wireguard configuration file that is served to clients
oidc_error.html: If a users login to the oidc provider as some issue (i.e user isnt registered for the device)
prompt_mfa_totp.html: Page for taking TOTP code entry
prompt_mfa_webauthn.html: Page for webauthn entry
qrcode_registration.html: When a client registers with the ?type=mobile option set, shows a QR code for the wireguard app on android/ios to simply registration
register_mfa_totp.html: Registration for TOTP that should show a QR code
register_mfa_webauth.html: Page to do webauthn registration
register_mfa.html: If multiple MFA methods are registered this page is displayed giving the user an option of what method to use
success.html: This page is not a template, and is displayed when a user is successfully authed, or if they attempt to access the authorisation endpoint while being authorised

cd internal/router
sudo go test -v .

Sudo is required to load the eBPF program into the kernel.

If you havent build the release docker image (used because it has a stable version of glibc) do the following:

cd release_builder
sudo docker build -t wag_builder .
cd ..

make docker

If you're looking to add your own features, or bug fixes to wag (thank you!). Please make sure that you've written a test for your changes if possible.
There are a few _test.go files around that give example on how to do this.

Then open a pull request and we can discuss it there.

If you like wag and use it to support your work flow, consider donating to the project. Your donations go directly towards the time and effort I put in, and the amount of support I can provide.

You can do this by either using the Support button on the side or the cryptocurrency wallets detailed below.

Monero (XMR):
8A8TRqsBKpMMabvt5RxMhCFWcuCSZqGV5L849XQndZB4bcbgkenH8KWJUXinYbF6ySGBznLsunrd1WA8YNPiejGp3FFfPND

Bitcoin (BTC):
bc1qm9e9sfrm7l7tnq982nrm6khnsfdlay07h0dxfr