debian-mirror-gitlab/doc/user/application_security/api_fuzzing/index.md
2021-04-17 20:07:23 +05:30

43 KiB

stage group info type
Secure Fuzz Testing To determine the technical writer assigned to the Stage/Group associated with this page, see https://about.gitlab.com/handbook/engineering/ux/technical-writing/#assignments reference, howto

Web API Fuzz Testing (ULTIMATE)

You can add web API fuzzing to your GitLab CI/CD pipelines. This helps you discover bugs and potential security issues that other QA processes may miss. API fuzzing performs fuzz testing of API operation parameters. Fuzz testing sets operation parameters to unexpected values in an effort to cause unexpected behavior and errors in the API backend.

We recommend that you use fuzz testing in addition to GitLab Secure's other security scanners and your own test processes. If you're using GitLab CI/CD, you can run fuzz tests as part your CI/CD workflow.

Requirements

  • One of the following web API types:
    • REST API
    • SOAP
    • GraphQL
    • Form bodies, JSON, or XML
  • One of the following assets to provide APIs to test:
    • OpenAPI v2 or v3 API definition
    • HTTP Archive (HAR) of API requests to test
    • Postman Collection v2.0 or v2.1

When fuzzing scans run

When using the API-Fuzzing.gitlab-ci.yml template, the fuzz job runs last, as shown here. To ensure API fuzzing scans the latest code, your CI pipeline should deploy changes to a test environment in one of the jobs preceding the fuzz job:

stages:
  - build
  - test
  - deploy
  - fuzz

Note that if your pipeline is configured to deploy to the same web server on each run, running a pipeline while another is still running could cause a race condition in which one pipeline overwrites the code from another. The API to scan should be excluded from changes for the duration of a fuzzing scan. The only changes to the API should be from the fuzzing scanner. Be aware that any changes made to the API (for example, by users, scheduled tasks, database changes, code changes, other pipelines, or other scanners) during a scan could cause inaccurate results.

Configuration

There are three ways to perform scans. See the configuration section for the one you wish to use:

Examples of both configurations can be found here:

WARNING: GitLab 14.0 will require that you place API fuzzing configuration files (for example, gitlab-api-fuzzing-config.yml) in your repository's .gitlab directory instead of your repository's root. You can continue using your existing configuration files as they are, but starting in GitLab 14.0, GitLab will not check your repository's root for configuration files.

Configuration form

WARNING: This feature might not be available to you. Check the version history note above for details.

The API fuzzing configuration form helps you create or modify your project's API fuzzing configuration. The form lets you choose values for the most common API fuzzing options and builds a YAML snippet that you can paste in your GitLab CI/CD configuration.

To generate an API Fuzzing configuration snippet:

  1. From your project's home page, go to Security & Compliance > Configuration in the left sidebar.
  2. Select Configure in the API Fuzzing row.
  3. Complete the form as needed. Read below for more information on available configuration options.
  4. Select Generate code snippet.

A modal opens with the YAML snippet corresponding to the options you've selected in the form.

API Fuzzing configuration snippet

Select Copy code and open .gitlab-ci.yml file to copy the snippet to your clipboard and be redirected to your project's .gitlab-ci.yml file where you can paste the YAML configuration.

Select Copy code only to copy the snippet to your clipboard and close the modal.

Enable or disable API Fuzzing configuration form (ULTIMATE)

The API Fuzzing configuration form is under development but ready for production use. It is deployed behind a feature flag that is enabled by default. GitLab administrators with access to the GitLab Rails console can opt to disable it.

To enable it:

Feature.enable(:api_fuzzing_configuration_ui)

To disable it:

Feature.disable(:api_fuzzing_configuration_ui)

OpenAPI Specification

Support for OpenAPI Specification v3 was introduced in GitLab 13.9.

The OpenAPI Specification (formerly the Swagger Specification) is an API description format for REST APIs. This section shows you how to configure API fuzzing by using an OpenAPI specification to provide information about the target API to test. OpenAPI specifications are provided as a file system resource or URL.

API fuzzing uses an OpenAPI document to generate the request body. When a request body is required, the body generation is limited to these body types:

  • application/x-www-form-urlencoded
  • multipart/form-data
  • application/json

Follow these steps to configure API fuzzing in GitLab with an OpenAPI specification:

  1. To use API fuzzing, you must include the API-Fuzzing.gitlab-ci.yml template that's provided as part of your GitLab installation. To do so, add the following to your .gitlab-ci.yml file:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
  2. Add the configuration file gitlab-api-fuzzing-config.yml to your repository's root as .gitlab-api-fuzzing.yml.

  3. The configuration file has several testing profiles defined with varying amounts of fuzzing. We recommend that you start with the Quick-10 profile. Testing with this profile completes quickly, allowing for easier configuration validation.

    Provide the profile by adding the FUZZAPI_PROFILE CI/CD variable to your .gitlab-ci.yml file, substituting Quick-10 for the profile you choose:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
    
  4. Provide the location of the OpenAPI specification. You can provide the specification as a file or URL. Specify the location by adding the FUZZAPI_OPENAPI variable:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
      FUZZAPI_OPENAPI: test-api-specification.json
    
  5. The target API instance's base URL is also required. Provide it by using the FUZZAPI_TARGET_URL variable or an environment_url.txt file.

    Adding the URL in an environment_url.txt file at your project's root is great for testing in dynamic environments. To run API fuzzing against an app dynamically created during a GitLab CI/CD pipeline, have the app persist its domain in an environment_url.txt file. API fuzzing automatically parses that file to find its scan target. You can see an example of this in our Auto DevOps CI YAML.

    Here's an example of using FUZZAPI_TARGET_URL:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
      FUZZAPI_OPENAPI: test-api-specification.json
      FUZZAPI_TARGET_URL: http://test-deployment/
    

This is a minimal configuration for API Fuzzing. From here you can:

WARNING: NEVER run fuzz testing against a production server. Not only can it perform any function that the API can, it may also trigger bugs in the API. This includes actions like modifying and deleting data. Only run fuzzing against a test server.

HTTP Archive (HAR)

The HTTP Archive format (HAR) is an archive file format for logging HTTP transactions. When used with the GitLab API fuzzer, HAR must contain records of calling the web API to test. The API fuzzer extracts all the requests and uses them to perform testing.

You can use various tools to generate HAR files:

WARNING: HAR files may contain sensitive information such as authentication tokens, API keys, and session cookies. We recommend that you review the HAR file contents before adding them to a repository.

Follow these steps to configure API fuzzing to use a HAR file that provides information about the target API to test:

  1. To use API fuzzing, you must include the API-Fuzzing.gitlab-ci.yml template that's provided as part of your GitLab installation. To do so, add the following to your .gitlab-ci.yml file:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
  2. Add the configuration file gitlab-api-fuzzing-config.yml to your repository's root as .gitlab-api-fuzzing.yml.

  3. The configuration file has several testing profiles defined with varying amounts of fuzzing. We recommend that you start with the Quick-10 profile. Testing with this profile completes quickly, allowing for easier configuration validation.

    Provide the profile by adding the FUZZAPI_PROFILE CI/CD variable to your .gitlab-ci.yml file, substituting Quick-10 for the profile you choose:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
    
  4. Provide the location of the HAR specification. You can provide the specification as a file or URL. URL support was introduced in GitLab 13.10 and later. Specify the location by adding the FUZZAPI_HAR variable:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
      FUZZAPI_HAR: test-api-recording.har
    
  5. The target API instance's base URL is also required. Provide it by using the FUZZAPI_TARGET_URL variable or an environment_url.txt file.

    Adding the URL in an environment_url.txt file at your project's root is great for testing in dynamic environments. To run API fuzzing against an app dynamically created during a GitLab CI/CD pipeline, have the app persist its domain in an environment_url.txt file. API fuzzing automatically parses that file to find its scan target. You can see an example of this in our Auto DevOps CI YAML.

    Here's an example of using FUZZAPI_TARGET_URL:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
      FUZZAPI_HAR: test-api-recording.har
      FUZZAPI_TARGET_URL: http://test-deployment/
    

This is a minimal configuration for API Fuzzing. From here you can:

WARNING: NEVER run fuzz testing against a production server. Not only can it perform any function that the API can, it may also trigger bugs in the API. This includes actions like modifying and deleting data. Only run fuzzing against a test server.

Postman Collection

The Postman API Client is a popular tool that developers and testers use to call various types of APIs. The API definitions can be exported as a Postman Collection file for use with API Fuzzing. When exporting, make sure to select a supported version of Postman Collection: v2.0 or v2.1.

When used with the GitLab API fuzzer, Postman Collections must contain definitions of the web API to test with valid data. The API fuzzer extracts all the API definitions and uses them to perform testing.

WARNING: Postman Collection files may contain sensitive information such as authentication tokens, API keys, and session cookies. We recommend that you review the Postman Collection file contents before adding them to a repository.

Follow these steps to configure API fuzzing to use a Postman Collection file that provides information about the target API to test:

  1. To use API fuzzing, you must include the API-Fuzzing.gitlab-ci.yml template that's provided as part of your GitLab installation. To do so, add the following to your .gitlab-ci.yml file:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
  2. Add the configuration file gitlab-api-fuzzing-config.yml to your repository's root as .gitlab-api-fuzzing.yml.

  3. The configuration file has several testing profiles defined with varying amounts of fuzzing. We recommend that you start with the Quick-10 profile. Testing with this profile completes quickly, allowing for easier configuration validation.

    Provide the profile by adding the FUZZAPI_PROFILE CI/CD variable to your .gitlab-ci.yml file, substituting Quick-10 for the profile you choose:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
    
  4. Provide the location of the Postman Collection specification. You can provide the specification as a file or URL. URL support was introduced in GitLab 13.10 and later. Specify the location by adding the FUZZAPI_POSTMAN_COLLECTION variable:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
      FUZZAPI_POSTMAN_COLLECTION: postman-collection_serviceA.json
    
  5. The target API instance's base URL is also required. Provide it by using the FUZZAPI_TARGET_URL variable or an environment_url.txt file.

    Adding the URL in an environment_url.txt file at your project's root is great for testing in dynamic environments. To run API fuzzing against an app dynamically created during a GitLab CI/CD pipeline, have the app persist its domain in an environment_url.txt file. API fuzzing automatically parses that file to find its scan target. You can see an example of this in our Auto DevOps CI YAML.

    Here's an example of using FUZZAPI_TARGET_URL:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
      FUZZAPI_POSTMAN_COLLECTION: postman-collection_serviceA.json
      FUZZAPI_TARGET_URL: http://test-deployment/
    

This is a minimal configuration for API Fuzzing. From here you can:

WARNING: NEVER run fuzz testing against a production server. Not only can it perform any function that the API can, it may also trigger bugs in the API. This includes actions like modifying and deleting data. Only run fuzzing against a test server.

Postman variables

Postman allows the developer to define placeholders that can be used in different parts of the requests. These placeholders are called variables, as explained in Using variables. You can use variables to store and reuse values in your requests and scripts. For example, you can edit the collection to add variables to the document:

Edit collection variable tab View

You can then use the variables in sections such as URL, headers, and others:

Edit request using variables View

Variables can be defined at different scopes (for example, Global, Collection, Environment, Local, and Data). In this example, they're defined at the Environment scope:

Edit environment variables View

When you export a Postman collection, only Postman collection variables are exported into the Postman file. For example, Postman does not export environment-scoped variables into the Postman file.

By default, the API fuzzer uses the Postman file to resolve Postman variable values. If a JSON file is set in a GitLab CI environment variable FUZZAPI_POSTMAN_COLLECTION_VARIABLES, then the JSON file takes precedence to get Postman variable values.

Although Postman can export environment variables into a JSON file, the format is not compatible with the JSON expected by FUZZAPI_POSTMAN_COLLECTION_VARIABLES.

Here is an example of using FUZZAPI_POSTMAN_COLLECTION_VARIABLES:

include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick-10
  FUZZAPI_POSTMAN_COLLECTION: postman-collection_serviceA.json
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_POSTMAN_COLLECTION_VARIABLES: variable-collection-dictionary.json

The file variable-collection-dictionary.json is a JSON document. This JSON is an object with key-value pairs for properties. The keys are the variables' names, and the values are the variables' values. For example:

{
   "base_url": "http://127.0.0.1/",
   "token": "Token 84816165151"
}

Authentication

Authentication is handled by providing the authentication token as a header or cookie. You can provide a script that performs an authentication flow or calculates the token.

HTTP Basic Authentication

HTTP basic authentication is an authentication method built in to the HTTP protocol and used in conjunction with transport layer security (TLS). To use HTTP basic authentication, two CI/CD variables are added to your .gitlab-ci.yml file:

  • FUZZAPI_HTTP_USERNAME: The username for authentication.
  • FUZZAPI_HTTP_PASSWORD: The password for authentication.

For the password, we recommended that you create a CI/CD variable (for example, TEST_API_PASSWORD) set to the password. You can create CI/CD variables from the GitLab projects page at Settings > CI/CD, in the Variables section. Use that variable as the value for FUZZAPI_HTTP_PASSWORD:

include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick-10
  FUZZAPI_HAR: test-api-recording.har
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_HTTP_USERNAME: testuser
  FUZZAPI_HTTP_PASSWORD: $TEST_API_PASSWORD

Bearer Tokens

Bearer tokens are used by several different authentication mechanisms, including OAuth2 and JSON Web Tokens (JWT). Bearer tokens are transmitted using the Authorization HTTP header. To use bearer tokens with API fuzzing, you need one of the following:

  • A token that doesn't expire
  • A way to generate a token that lasts the length of testing
  • A Python script that API fuzzing can call to generate the token
Token doesn't expire

If the bearer token doesn't expire, use the FUZZAPI_OVERRIDES_ENV variable to provide it. This variable's content is a JSON snippet that provides headers and cookies to add to API fuzzing's outgoing HTTP requests.

Follow these steps to provide the bearer token with FUZZAPI_OVERRIDES_ENV:

  1. Create a CI/CD variable, for example TEST_API_BEARERAUTH, with the value {"headers":{"Authorization":"Bearer dXNlcm5hbWU6cGFzc3dvcmQ="}} (substitute your token). You can create CI/CD variables from the GitLab projects page at Settings > CI/CD, in the Variables section.

  2. In your .gitlab-ci.yml file, set FUZZAPI_OVERRIDES_ENV to the variable you just created:

    include:
      - template: API-Fuzzing.gitlab-ci.yml
    
    variables:
      FUZZAPI_PROFILE: Quick-10
      FUZZAPI_OPENAPI: test-api-specification.json
      FUZZAPI_TARGET_URL: http://test-deployment/
      FUZZAPI_OVERRIDES_ENV: $TEST_API_BEARERAUTH
    
  3. To validate that authentication is working, run an API fuzzing test and review the fuzzing logs and the test API's application logs.

Token generated at test runtime

If the bearer token must be generated and doesn't expire during testing, you can provide to API fuzzing a file containing the token. A prior stage and job, or part of the API fuzzing job, can generate this file.

API fuzzing expects to receive a JSON file with the following structure:

{
  "headers" : {
    "Authorization" : "Bearer dXNlcm5hbWU6cGFzc3dvcmQ="
  }
}

This file can be generated by a prior stage and provided to API fuzzing through the FUZZAPI_OVERRIDES_FILE CI/CD variable.

Set FUZZAPI_OVERRIDES_FILE in your .gitlab-ci.yml file:

include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick
  FUZZAPI_OPENAPI: test-api-specification.json
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_OVERRIDES_FILE: output/api-fuzzing-overrides.json

To validate that authentication is working, run an API fuzzing test and review the fuzzing logs and the test API's application logs.

Token has short expiration

If the bearer token must be generated and expires prior to the scan's completion, you can provide a program or script for the API fuzzer to execute on a provided interval. The provided script runs in an Alpine Linux container that has Python 3 and Bash installed. If the Python script requires additional packages, it must detect this and install the packages at runtime.

The script must create a JSON file containing the bearer token in a specific format:

{
  "headers" : {
    "Authorization" : "Bearer dXNlcm5hbWU6cGFzc3dvcmQ="
  }
}

You must provide three CI/CD variables, each set for correct operation:

  • FUZZAPI_OVERRIDES_FILE: JSON file the provided command generates.
  • FUZZAPI_OVERRIDES_CMD: Command that generates the JSON file.
  • FUZZAPI_OVERRIDES_INTERVAL: Interval (in seconds) to run command.

For example:

include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick-10
  FUZZAPI_OPENAPI: test-api-specification.json
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_OVERRIDES_FILE: output/api-fuzzing-overrides.json
  FUZZAPI_OVERRIDES_CMD: renew_token.py
  FUZZAPI_OVERRIDES_INTERVAL: 300

To validate that authentication is working, run an API fuzzing test and review the fuzzing logs and the test API's application logs.

Configuration files

To get you started quickly, GitLab provides the configuration file gitlab-api-fuzzing-config.yml. This file has several testing profiles that perform various numbers of tests. The run time of each profile increases as the test numbers go up. To use a configuration file, add it to your repository's root as .gitlab-api-fuzzing.yml.

Profile Fuzz Tests (per parameter)
Quick-10 10
Medium-20 20
Medium-50 50
Long-100 100

Available CI/CD variables

CI/CD variable Description
FUZZAPI_VERSION Specify API Fuzzing container version. Defaults to latest.
FUZZAPI_TARGET_URL Base URL of API testing target.
FUZZAPI_CONFIG API Fuzzing configuration file. Defaults to .gitlab-apifuzzer.yml.
FUZZAPI_PROFILE Configuration profile to use during testing. Defaults to Quick.
FUZZAPI_REPORT Scan report filename. Defaults to gl-api_fuzzing-report.xml.
FUZZAPI_OPENAPI OpenAPI specification file or URL.
FUZZAPI_HAR HTTP Archive (HAR) file.
FUZZAPI_POSTMAN_COLLECTION Postman Collection file.
FUZZAPI_POSTMAN_COLLECTION_VARIABLES Path to a JSON file to extract postman variable values.
FUZZAPI_OVERRIDES_FILE Path to a JSON file containing overrides.
FUZZAPI_OVERRIDES_ENV JSON string containing headers to override.
FUZZAPI_OVERRIDES_CMD Overrides command.
FUZZAPI_OVERRIDES_INTERVAL How often to run overrides command in seconds. Defaults to 0 (once).
FUZZAPI_HTTP_USERNAME Username for HTTP authentication.
FUZZAPI_HTTP_PASSWORD Password for HTTP authentication.

Overrides

API Fuzzing provides a method to add or override specific items in your request, for example:

  • Headers
  • Cookies
  • Query string
  • Form data
  • JSON nodes
  • XML nodes

You can use this to inject semantic version headers, authentication, and so on. The authentication section includes examples of using overrides for that purpose.

Overrides use a JSON document, where each type of override is represented by a JSON object:

{
  "headers": {
    "header1": "value",
    "header2": "value"
  },
  "cookies": {
    "cookie1": "value",
    "cookie2": "value"
  },
  "query":      {
    "query-string1": "value",
    "query-string2": "value"
  },
  "body-form":  {
    "form-param1": "value",
    "form-param1": "value",
  },
  "body-json":  {
    "json-path1": "value",
    "json-path2": "value",
  },
  "body-xml" :  {
    "xpath1":    "value",
    "xpath2":    "value",
  }
}

Example of setting a single header:

{
  "headers": {
    "Authorization": "Bearer dXNlcm5hbWU6cGFzc3dvcmQ="
  }
}

Example of setting both a header and cookie:

{
  "headers": {
    "Authorization": "Bearer dXNlcm5hbWU6cGFzc3dvcmQ="
  },
  "cookies": {
    "flags": "677"
  }
}

Example usage for setting a body-form override:

{
  "body-form":  {
    "username": "john.doe"
  }
}

The override engine uses body-form when the request body has only form-data content.

Example usage for setting a body-json override:

{
  "body-json":  {
    "$.credentials.access-token": "iddqd!42.$"
  }
}

Note that each JSON property name in the object body-json is set to a JSON Path expression. The JSON Path expression $.credentials.access-token identifies the node to be overridden with the value iddqd!42.$. The override engine uses body-json when the request body has only JSON content.

For example, if the body is set to the following JSON:

{
    "credentials" : {
        "username" :"john.doe",
        "access-token" : "non-valid-password"
    }
}

It is changed to:

{
    "credentials" : {
        "username" :"john.doe",
        "access-token" : "iddqd!42.$"
    }
}

Here's an example for setting a body-xml override. The first entry overrides an XML attribute and the second entry overrides an XML element:

{
  "body-xml" :  {
    "/credentials/@isEnabled": "true",
    "/credentials/access-token/text()" : "iddqd!42.$"
  }
}

Note that each JSON property name in the object body-xml is set to an XPath v2 expression. The XPath expression /credentials/@isEnabled identifies the attribute node to override with the value true. The XPath expression /credentials/access-token/text() identifies the element node to override with the value iddqd!42.$. The override engine uses body-xml when the request body has only XML content.

For example, if the body is set to the following XML:

<credentials isEnabled="false">
  <username>john.doe</username>
  <access-token>non-valid-password</access-token>
</credentials>

It is changed to:

<credentials isEnabled="true">
  <username>john.doe</username>
  <access-token>iddqd!42.$</access-token>
</credentials>

You can provide this JSON document as a file or environment variable. You may also provide a command to generate the JSON document. The command can run at intervals to support values that expire.

Using a file

To provide the overrides JSON as a file, the FUZZAPI_OVERRIDES_FILE CI/CD variable is set. The path is relative to the job current working directory.

Here's an example .gitlab-ci.yml:

include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick
  FUZZAPI_OPENAPI: test-api-specification.json
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_OVERRIDES_FILE: output/api-fuzzing-overrides.json

Using a CI/CD variable

To provide the overrides JSON as a CI/CD variable, use the FUZZAPI_OVERRIDES_ENV variable. This allows you to place the JSON as variables that can be masked and protected.

In this example .gitlab-ci.yml, the FUZZAPI_OVERRIDES_ENV variable is set directly to the JSON:

include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick
  FUZZAPI_OPENAPI: test-api-specification.json
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_OVERRIDES_ENV: '{"headers":{"X-API-Version":"2"}}'

In this example .gitlab-ci.yml, the SECRET_OVERRIDES variable provides the JSON. This is a group or instance level CI/CD variable defined in the UI:

include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick
  FUZZAPI_OPENAPI: test-api-specification.json
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_OVERRIDES_ENV: $SECRET_OVERRIDES

Using a command

If the value must be generated or regenerated on expiration, you can provide a program or script for the API fuzzer to execute on a specified interval. The provided script runs in an Alpine Linux container that has Python 3 and Bash installed. If the Python script requires additional packages, it must detect this and install the packages at runtime. The script creates the overrides JSON file as defined above.

You must provide three CI/CD variables, each set for correct operation:

  • FUZZAPI_OVERRIDES_FILE: File generated by the provided command.
  • FUZZAPI_OVERRIDES_CMD: Command to generate JSON file.
  • FUZZAPI_OVERRIDES_INTERVAL: Interval in seconds to run command.
include:
  - template: API-Fuzzing.gitlab-ci.yml

variables:
  FUZZAPI_PROFILE: Quick
  FUZZAPI_OPENAPI: test-api-specification.json
  FUZZAPI_TARGET_URL: http://test-deployment/
  FUZZAPI_OVERRIDES_FILE: output/api-fuzzing-overrides.json
  FUZZAPI_OVERRIDES_CMD: renew_token.py
  FUZZAPI_OVERRIDES_INTERVAL: 300

Header Fuzzing

Header fuzzing is disabled by default due to the high number of false positives that occur with many technology stacks. When header fuzzing is enabled, you must specify a list of headers to include in fuzzing.

Each profile in the default configuration file has an entry for GeneralFuzzingCheck. This check performs header fuzzing. Under the Configuration section, you must change the HeaderFuzzing and Headers settings to enable header fuzzing.

This snippet shows the Quick-10 profile's default configuration with header fuzzing disabled:

- Name: Quick-10
  DefaultProfile: Empty
  Routes:
  - Route: *Route0
    Checks:
    - Name: FormBodyFuzzingCheck
      Configuration:
        FuzzingCount: 10
        UnicodeFuzzing: true
    - Name: GeneralFuzzingCheck
      Configuration:
        FuzzingCount: 10
        UnicodeFuzzing: true
        HeaderFuzzing: false
        Headers:
    - Name: JsonFuzzingCheck
      Configuration:
        FuzzingCount: 10
        UnicodeFuzzing: true
    - Name: XmlFuzzingCheck
      Configuration:
        FuzzingCount: 10
        UnicodeFuzzing: true

HeaderFuzzing is a boolean that turns header fuzzing on and off. The default setting is false for off. To turn header fuzzing on, change this setting to true:

    - Name: GeneralFuzzingCheck
      Configuration:
        FuzzingCount: 10
        UnicodeFuzzing: true
        HeaderFuzzing: true
        Headers:

Headers is a list of headers to fuzz. Only headers listed are fuzzed. To fuzz a header used by your APIs, add an entry for it using the syntax - Name: HeaderName. For example, to fuzz a custom header X-Custom, add - Name: X-Custom:

    - Name: GeneralFuzzingCheck
      Configuration:
        FuzzingCount: 10
        UnicodeFuzzing: true
        HeaderFuzzing: true
        Headers:
          - Name: X-Custom

You now have a configuration to fuzz the header X-Custom. Use the same notation to list additional headers:

    - Name: GeneralFuzzingCheck
      Configuration:
        FuzzingCount: 10
        UnicodeFuzzing: true
        HeaderFuzzing: true
        Headers:
          - Name: X-Custom
          - Name: X-AnotherHeader

Repeat this configuration for each profile as needed.

Running your first scan

When configured correctly, a CI/CD pipeline contains a fuzz stage and an apifuzzer_fuzz or apifuzzer_fuzz_dnd job. The job only fails when an invalid configuration is provided. During normal operation, the job always succeeds even if faults are identified during fuzz testing.

Faults are displayed on the Security pipeline tab with the suite name. When testing against the repositories default branch, the fuzzing faults are also shown on the Security & Compliance's Vulnerability Report page.

To prevent an excessive number of reported faults, the API fuzzing scanner limits the number of faults it reports.

Viewing fuzzing faults

The API Fuzzing analyzer produces a JSON report that is collected and used to populate the faults into GitLab vulnerability screens. Fuzzing faults show up as vulnerabilities with a severity of Unknown.

The faults that API fuzzing finds require manual investigation and aren't associated with a specific vulnerability type. They require investigation to determine if they are a security issue, and if they should be fixed. See handling false positives for information about configuration changes you can make to limit the number of false positives reported.

For additional information, see View details of an API Fuzzing vulnerability.

Security Dashboard

Fuzzing faults show up as vulnerabilities with a severity of Unknown. The Security Dashboard is a good place to get an overview of all the security vulnerabilities in your groups, projects and pipelines. For more information, see the Security Dashboard documentation.

Interacting with the vulnerabilities

Fuzzing faults show up as vulnerabilities with a severity of Unknown. Once a fault is found, you can interact with it. Read more on how to address the vulnerabilities.

Handling False Positives

False positives can be handled in two ways:

  • Turn off the Check producing the false positive. This prevents the check from generating any faults. Example checks are the JSON Fuzzing Check, and Form Body Fuzzing Check.
  • Fuzzing checks have several methods of detecting when a fault is identified, called Asserts. Asserts can also be turned off and configured. For example, the API fuzzer by default uses HTTP status codes to help identify when something is a real issue. If an API returns a 500 error during testing, this creates a fault. This isn't always desired, as some frameworks return 500 errors often.

Turn off a Check

Checks perform testing of a specific type and can be turned on and off for specific configuration profiles. The provided configuration files define several profiles that you can use. The profile definition in the configuration file lists all the checks that are active during a scan. To turn off a specific check, simply remove it from the profile definition in the configuration file. The profiles are defined in the Profiles section of the configuration file.

Example profile definition:

Profiles:
  - Name: Quick-10
    DefaultProfile: Quick
    Routes:
      - Route: *Route0
        Checks:
          - Name: FormBodyFuzzingCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true
          - Name: GeneralFuzzingCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true
          - Name: JsonFuzzingCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true
          - Name: XmlFuzzingCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true

To turn off the General Fuzzing Check you can remove these lines:

- Name: GeneralFuzzingCheck
  Configuration:
    FuzzingCount: 10
    UnicodeFuzzing: true

This results in the following YAML:

- Name: Quick-10
  DefaultProfile: Quick
  Routes:
    - Route: *Route0
      Checks:
        - Name: FormBodyFuzzingCheck
          Configuration:
            FuzzingCount: 10
            UnicodeFuzzing: true
        - Name: JsonFuzzingCheck
          Configuration:
            FuzzingCount: 10
            UnicodeFuzzing: true
        - Name: XmlFuzzingCheck
          Configuration:
            FuzzingCount: 10
            UnicodeFuzzing: true

Turn off an Assertion for a Check

Assertions detect faults in tests produced by checks. Many checks support multiple Assertions such as Log Analysis, Response Analysis, and Status Code. When a fault is found, the Assertion used is provided. To identify which Assertions are on by default, see the Checks default configuration in the configuration file. The section is called Checks.

This example shows the FormBody Fuzzing Check:

Checks:
  - Name: FormBodyFuzzingCheck
    Configuration:
      FuzzingCount: 30
      UnicodeFuzzing: true
    Assertions:
      - Name: LogAnalysisAssertion
      - Name: ResponseAnalysisAssertion
      - Name: StatusCodeAssertion

Here you can see three Assertions are on by default. A common source of false positives is StatusCodeAssertion. To turn it off, modify its configuration in the Profiles section. This example provides only the other two Assertions (LogAnalysisAssertion, ResponseAnalysisAssertion). This prevents FormBodyFuzzingCheck from using StatusCodeAssertion:

Profiles:
  - Name: Quick-10
    DefaultProfile: Quick
    Routes:
      - Route: *Route0
        Checks:
          - Name: FormBodyFuzzingCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true
            Assertions:
              - Name: LogAnalysisAssertion
              - Name: ResponseAnalysisAssertion
          - Name: GeneralFuzzingCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true
          - Name: JsonFuzzingCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true
          - Name: XmlInjectionCheck
            Configuration:
              FuzzingCount: 10
              UnicodeFuzzing: true

Glossary

  • Assert: Assertions are detection modules used by checks to trigger a fault. Many assertions have configurations. A check can use multiple Assertions. For example, Log Analysis, Response Analysis, and Status Code are common Assertions used together by checks. Checks with multiple Assertions allow them to be turned on and off.
  • Check: Performs a specific type of test, or performed a check for a type of vulnerability. For example, the JSON Fuzzing Check performs fuzz testing of JSON payloads. The API fuzzer is comprised of several checks. Checks can be turned on and off in a profile.
  • Fault: During fuzzing, a failure identified by an Assert is called a fault. Faults are investigated to determine if they are a security vulnerability, a non-security issue, or a false positive. Faults don't have a known vulnerability type until they are investigated. Example vulnerability types are SQL Injection and Denial of Service.
  • Profile: A configuration file has one or more testing profiles, or sub-configurations. You may have a profile for feature branches and another with extra testing for a main branch.