ctf_proxy creates a proxy process for each configured service, with the possibility to block malicious attacks by means of custom Python filters.
This tool is purposely made for Attack/Defense CTF competitions.
- Dropping or modifying TCP and HTTP messages through simple Python filters
- Access to the previously exchanged messages over the same TCP connection for stateful filtering
- A configurable keyword gets sent as response to attacks to easily highlight and analyze them in PCAP files
- HTTP parsing inside filters for easy access to headers and parameters
- Filters are automatically reloaded on the fly when modified
- Log by printing to stdout inside filters
- Support for SSL/TLS services by providing certificates and keys
- Support for IPv6
- Easy Docker configuration
- Proxy failover mechanism using NGINX
- DoS: maintain attackers' sockets alive by periodically sending bytes to massively slow down their scripts if they don't expect it
This architecture implements a failover mechanism through the usage of NGINX:
- ctf_proxy is up and running: all traffic will flow through it, filtering every attack coming in the way
- ctf_proxy is down: the proxy will be bypassed to ensure maximum service availability throughout the competition.
You can configure each service to be proxied using proxy/config/config.json.
{
"services": [
{
"name": "generic_service",
"target_ip": "10.10.10.10",
"target_port": 80,
"listen_port": 80
},
{
"name": "ssl_service",
"target_ip": "localhost",
"target_port": 443,
"listen_port": 500,
"http": true,
"ssl": {
"server_certificate": "server.pem",
"server_key": "server.pem",
"client_certificate": "client_crt.pem",
"client_key": "client_key.pem",
"ca_file": "ca_file.pem"
}
}
],
"global_config": {
"keyword": "KEYWORD FOR PACKET ANALYZERS",
"verbose": false,
"nginx": {
"connect_timeout": 5,
"max_fails": 1,
"fail_timeout": 20
},
"dos": {
"enabled": true,
"duration": 60,
"interval": 2
},
"max_stored_messages": 10,
"max_message_size": 65535
}
}In the services list, the following parameters can be set for each service:
- name: name that will be used for logging and filter modules generation
- target_ip: service IP/hostname
- target_port: service port
- listen_port: proxy port to listen on
- http: (optional) must be set to
trueto enable HTTP parsing, default=False - ssl: (optional, required for SSL services) set SSL files for secure communication. Each file will be looked in the
proxy/config/certificatesfolder:- server_certificate: server certificate
- server_key: server key file
- client_certificate: (optional) for client authentication between proxy and service
- client_key: (optional) for client authentication between proxy and service
- ca_file: (optional) certificate authority for checking client authentication between client and proxy
The global_config contains:
- keyword: string to be sent as response to malicious packets, to facilitate packet inspections
- verbose: verbose mode
- nginx: failover configurations:
- connect_timeout: (seconds) timeout given to NGINX to declare the proxy as down when trying a connection
- fail_timeout: (seconds) sets the time during which the specified number of unsuccessful attempts to communicate with the server should happen to consider the server unavailable and the period of time the server will be considered unavailable.
- max_fails: (seconds) sets the number of unsuccessful attempts to communicate with the server that should happen in the duration set by the
fail_timeoutparameter to consider the server unavailable.
- dos: slow down attackers by keeping the socket alive:
- enabled: (boolean)
- duration: (seconds) time to wait before closing the socket
- interval: (seconds) time interval between packets to keep the socket alive
- max_stored_messages: maximum number of stored messages in the
previous_messagesqueue in the Stream objects - max_message_size: maximum message size after which the message will be truncated before being stored in the
previous_messagesqueue in the Stream objects
Clone the repository:
git clone https://github.com/ByteLeMani/ctf_proxy
cd ctf_proxy
Edit the docker-compose.yml file with the current port mapping. For example, if you want to listen on port 8080 and 9090, you need to set:
ports:
- 8080:8080
- 9090:9090Edit the proxy configuration file at proxy/config/config.json.
Run the container:
docker compose up --build -dYou may want to add this configuration to the services docker-compose.yml file and remove any address/port binding it may have, to take advantage of the docker network DNS.
networks:
default:
name: ctf_network
external: trueThis way, you can use the services' hostname directly in the target_ip parameters. Moreover, as the services are connected to the proxy network they are reachable inside it without exposing or changing any port, but not reachable from the outside.
Note: the network must already exist, so the proxy should be run before the services.
All the above configuration can be done with the provided setup script. Either clone the repository, install ruamel.yaml and run the script or directly run
pip install ruamel.yaml; curl -s https://raw.githubusercontent.com/ByteLeMani/ctf_proxy/main/setup_proxy.py > setup_proxy.py; python3 setup_proxy.py
The script will work as long as it is called from inside ctf_proxy or besides all the services. In the latter case it will clone ctf_proxy before configuring it.
You can manually provide paths for the services as arguments when you call the script. If you don't it will automatically scan cwd and look for services, asking for confirmation.
python3 setup_proxy.py /path/to/service1 /path/to/service2 ...
These are the core filtering entities of the proxy. For each proxied service, a pair of modules is automatically generated inside the proxy/filter_modules/<service_name>/ folder the first time you run the proxy. Modules execution will follow this flow:
Inside the modules you will find an execute method that receives a Stream object from the proxy and returns to the proxy whether the stream contains an attack or not. If an attack is found, the proxy will send to the attacker a custom string (KEYWORD + " " + SERVICE NAME + " " + ATTACK NAME) to easily find attacks in PCAP files if a packet analyzer is used in the system. At this point, the socket will be kept alive to interfere with attackers' scripts if the dos parameter is enabled, otherwise the socket will be closed.
dos may not work properly if ssl is enabled
Stream is the interface for TCPStream and HTTPStream classes and contains the current and previous exchanged messages on the socket.
TCPStream is used in case of TCP only connections. The previous_messages queue contains the latest max_stored_messages messages that went through the socket before the current received message. The latter will be stored in the current_message variable and can be modified to alter the content of the message that will be sent from the proxy. Note that in this case you should return False inside the filter to not drop the connection.
HTTPStream is used in case of HTTP connections. Two more variables are available: previous_http_messages and current_http_message. They are the parsed version of the correspective raw variables as HttpMessage objects. The actual values sent through the socket are the ones stored in current_message, which means you should edit this variable to alter the content of the message sent from the proxy.
Content-Length header in the raw HTTP message if you intend to modify it.
To add a new filter, define a new function inside the class Module called as the name of the attack (or a custom one if you prefer) that accepts a Stream object as parameter and returns a boolean (True if attack found, False if not). You will find a filter example in proxy/filter_modules/template.py.
Every module will be automatically reloaded on the fly by simply modifying it. If an exception is thrown during the import, the module will not be loaded and the previous version will be used instead. If an exception is thrown at runtime, the packet will simply flow through the proxy.
For stateful filters, you can build and use the local Mongo database. You can access the database inside the modules through the DBManager interface. You can find some examples in proxy/filter_modules/example_functions.py.
A simple log file proxy/log.txt will count all the blocked packets for each service. If the file already exists at startup, the proxy will update the counts based on their initial value inside the file. Otherwise, the file will be created at startup.
You can monitor the file by using:
watch cat log.txt
Any relevant information will be printed to stdout. Set verbose to true for more info.
If you're using the Docker version, you can inspect the docker logs to access them:
docker logs --follow proxy
Written by:
This tool is inspired by and partially based on the TCP proxy logic of tcpproxy.py.