OWASP Threat Dragon

Threat Dragon is a free, open-source, cross-platform threat modelling application including system diagramming and a rule engine to auto-generate threats/mitigations. It is an OWASP Incubator Project. The focus of the project is on great UX, a powerful rule engine and integration with other development lifecycle tools.

The application comes in two variants:

1. A web application (this repo): For the web application, models files are stored in GitHub (other storage will become available). We are currently maintaining a working protoype in sych with the master code branch.

2. An desktop application: This is based on Electon. There are builds available for Windows and OSX (Linux will follow at some point). For this variant, models arfe stored on the local filesystem.

End user help is available for both variants.

This repository contains the files for the web application variant.

Core files that are shared between both thbe desktop and web variants stored in an seperate repo and are installable as a seperate package.

Installing

ThreatDragon is a Single Page Application (SPA) using Angular on the client and node.js on the server. To build and run locally, follow these steps:

Install Git and node.js. To get the code, go to where you want your code to be located and do

git init

git clone https://github.com/mike-goodwin/owasp-threat-dragon.git

This installs code in two sub-folders. One for the main application (td) and one for the unit tests (td.tests). To install, do:

npm install

Environment variables

Threat Dragon uses GitHub to store threat models, so you need to go to your GitHub account and register it as a GitHub application. Once you have done that you need to set the Client ID and Client Secret as environment variables (GITHUB_CLIENT_ID and GITHUB_CLIENT_SECRET).

You also need to set a session signing key environment variable (SESSION_SIGNING_KEY).

Once a user is signed in, their session information contains an OAuth access token with write access to their GitHub repos. For security, this is encrypted before storage in the session. The session encryption supports multiple keys so that they can be expired without any interruption to the running application. The primary key is always used for encryption. Retired keys can be kept available for decrypting existing sessions. Once all sessions are using the new primary key (typically this will be around 60 minutes maximum), the old one can be safely removed. The keys are stored as a JSON string in the SESSION_ENCRYPTION_KEYS environment variable. For example:

[{\"isPrimary\": true, \"id\": 0, \"value\": \"abcdef\"}, {\"isPrimary\": false, \"id\": 1, \"value\": \"ghijkl\"}]

If you are developing locally, you can choose to store the session data in memory using the express-session in-memory store. To do this the SESSION_STOREenvironment variale to local. As mentioned in the express-session docs this is for development only - it is not suitable for production. To remind you of this, Threat Dragon will write a log message at severity ERROR when it starts if the in memory session store is used.

For production use, Threat Dragon currently supports Azure Table Storage for the session store via connect-azuretables. To make this store work you need to specify an Azure Storage Account and key as environment variables AZURE_STORAGE_ACCOUNT and AZURE_STORAGE_ACCESS_KEY. See the connect-azuretables documentation for more options.

If you want to use an alternative session store in production, install it and edit the session.config.js file.

Lastly, by default, Threat Dragon will set the secure flag on cookies. To override this for development purposes, set the NODE_ENV environment variable to development.

Running the application

Once your environment variables are set up, start the node web server:

npm start

If you then browse to http://localhost:3000 you should see the running application.

Building

The basic build script is:

npm run build

See package.json for other build tasks.

Running the unit tests

The unit tests are written using Jasmine and Karma. Coverage is by Istanbul. A few different npm tasks are available:

• pretest: runs jshint without the unit tests
• test-client-phantomjs, test-client-firefox, test-client-chrome, test-client-ie: runs client side tests using the specified browser
• test-server: runs the server side tests
• test: runs jshint, client side tests on Firefox and PhantomJS and server side tests (this is what runs on Travis CI)
• test-local: runs jshint, client side tests on all browsers and then the server side tests (useful as a pre-push git hook)
• citest: continously runs client side tests in PhantomJS with --single-run false (useful while coding)

Note: If you are on Windows and are having problems installing Karma, the simplest way to resolve this seems to be to install Python v2.7.x (not v3+) and then install Visual Studio Express as per the SO answer suggested in this link. This sounds mad, but the alternative is a world of pain installing various patches and components one by one. At least it's free :o/

Contributing

PRs, feature requests, bug reports and feedback of any kind are very welcome. We are trying to keep the test coverage relatively high, so please try to include tests in any PRs and make PRs on the development branch.

Project leader

Mike Goodwin ([email protected])