Skip to content

CiscoSecurity/tr-05-serverless-have-i-been-pwned

Repository files navigation

Gitter Chat

NOTE! This code has been upgraded and the current release no longer supports installation in AWS

If you wish to deploy in AWS, use this previous release.

Have I Been Pwned Relay (Cisco Hosted)

A Cisco SecureX Concrete Relay implementation using Have I Been Pwned as a third-party Cyber Threat Intelligence service provider.

The Relay itself is just a simple application written in Python that can be easily packaged and deployed. This relay is now Cisco Hosted and no longer requires AWS Lambda.

The code is provided here purely for educational purposes.

Rationale

  • We need an application that will translate API requests from SecureX Threat Response to the third-party integration, and vice versa.
  • We need an application that can be completely self contained within a virtualized container using Docker.

Testing (Optional)

Open the code folder in your terminal.

cd code

If you want to test the application you will require Docker and several dependencies from the Pipfile file:

pip install --no-cache-dir --upgrade pipenv && pipenv install --dev

You can perform two kinds of testing:

  • Run static code analysis checking for any semantic discrepancies and PEP 8 compliance:

    flake8 .

  • Run the suite of unit tests and measure the code coverage: coverage run --source api/ -m pytest --verbose tests/unit/ && coverage report

NOTE. If you need input data for testing purposes you can use data from the observables.json file.

Building the Docker Container

In order to build the application, we need to use a Dockerfile.

  1. Open a terminal. Build the container image using the docker build command.
docker build -t tr-05-have-i-been-pwned .
  1. Once the container is built, and an image is successfully created, start your container using the docker run command and specify the name of the image we have just created. By default, the container will listen for HTTP requests using port 9090.
docker run -dp 9090:9090 --name tr-05-have-i-been-pwned tr-05-have-i-been-pwned
  1. Watch the container logs to ensure it starts correctly.
docker logs tr-05-have-i-been-pwned
  1. Once the container has started correctly, open your web browser to http://localhost:9090. You should see a response from the container.
curl http://localhost:9090

Implementation Details

This application was developed and tested under Python version 3.9.

Implemented Relay Endpoints

  • POST /health

    • Verifies the Authorization Bearer JWT and decodes it to restore the original credentials.
    • Authenticates to the underlying external service to check that the provided credentials are valid and the service is available at the moment.
  • POST /observe/observables

    • Accepts a list of observables and filters out unsupported ones.
    • Verifies the Authorization Bearer JWT and decodes it to restore the original credentials.
    • Makes a series of requests to the underlying external service to query for some cyber threat intelligence data on each supported observable.
    • Maps the fetched data into appropriate CTIM entities.
    • Returns a list per each of the following CTIM entities (if any extracted):
      • Indicator,
      • Sighting,
      • Relationship.
  • POST /refer/observables

    • Accepts a list of observables and filters out unsupported ones.
    • Builds a search link per each supported observable to pivot back to the underlying external service and look up the observable there.
    • Returns a list of those links.
  • POST /version

    • Returns the current version of the application.

Supported Types of Observables

  • email

CTIM Mapping Specifics

Each HIBP breach for an email generates 3 CTIM entities: an Indicator, a Sighting, and the corresponding Relationship between them. The actual mapping from HIBP fields to CTIM fields is quite straightforward.

The only non-obvious piece of the mapping is the logic for inferring the actual values for the confidence and severity fields. Suppose there is a breach for an email. If the breach is verified (field IsVerified, type boolean), then the value for confidence will be High, otherwise Medium. At the same time, each breach also knows some information about the nature of the data compromised in the breach as a string array of impacted data classes (field DataClasses, type string[]). Thus, if the breach is verified and the password is also known to be compromised (i.e. the data classes contain the Passwords data class), then the value for severity will be High, otherwise Medium.

The rules mentioned above can be easily expressed in Python using the following code snippet:

entity['confidence'] = ['Medium', 'High'][breach['IsVerified']]

entity['severity'] = ['Medium', 'High'][
    breach['IsVerified'] and 'Passwords' in breach['DataClasses']
]

(the entity here is either an Indicator or a Sighting).