Combatting Self-XSS (Part 2)

The immediate context to this post is Self-XSS but it may have a wider context in avoiding other Internet scams.

I’d like to frame Self-XSS in terms of a human script execution engine. It’s an interpreter - and probably not JITed, but it works with a broad and poorly specified grammar. Not all humans have the same interpreter, but there’s enough of a dominant language to provide a mono-culture which makes attack easy.

However not all script execution engines are created equal. Some come with advanced scam detection routines. Some have random timeouts and many have surprising execution errors. It’s not my intention to be derogatory, but it’s a fact that not everyone has English skills the same.

So Self-XSS is a script that uses this script execution engine to coerce someone to do the attacker’s bidding, probably to their own detriment. If you're writing a scam script, you need to keep the scripts simple enough to avoid the victims’ scam-detection, timeouts and errors.

These problems mean that we might be able to draw a graph of script complexity vs completion rate (where ‘completion’ is a scam that succeeded). It might look like this:

Graph of complexity vs completion rate showing exponential drop off

I’d like to suggest that humans are not that simple and that we shouldn’t expect something as predictable as that.

Many (most?) of the tasks that we complete on computers can be done purely using unconcious / heuristic / system 1 processing (in dual process) terms. That is to say using parts of the brain that are fast, pattern-based and non-analytical. Its less common that we need to use concious / systematic / system 2 processing (slow, lazy, analytical). (“Thinking, Fast and Slow” is probably the canonical work to understand these terms in more depth)

I’d suggest that, for most people, scam detection is a system 2 operation. So forcing use of system 2 to prevent scams might help. (Obviously this is at odds with the “don’t make me think” approach to good UX design - I’m obviously not arguing that bad design is a good thing).

We don’t have a lot of data for Self-XSS. The data that I know of is:

Attack Completion Rate Complexity
Facebook <10% 3 steps
Win+R [1] <1% 5+ steps
Other attacks [2] <1% 4+ steps

It seems reasonable that an infinitely simple script would always succeed, and that at some point of complexity the completion rate will drop to 0. So that something like 5 data points.

It would be fair to criticise this as ‘not much data’, which is part of the point:

  • If you have more data in places where Self-XSS attacks have worked, please publish them and tell me.
  • If you know of research into the use of system 1 vs. system 2 in detecting scams, I’d love to hear about it.

However, given the lack of data, it seems to me that there is a surprising drop-off in completion rate with increasing script complexity. There isn’t much any evidence for attacks that were somewhat complex, and got some people - it looks as though all attacks that were somewhat complex were surprisingly unsuccessful.

So the drop-off rate might look something a bit more like this:

Graph of complexity vs completion rate showing more sudden drop-off than before

Which might in some small way indicate a point in script complexity where system 2 needed to get involved, and where the analytical part of system 2 said “Wait, am I being scammed?”

Summary: We shouldn't think of scams purely in terms of smart people not getting scammed, and some sort of level of intellegence providing scam protection. Perhaps for many people, scams are detected when system 2 gets involved, and there may be ways in which we can design software to protect people.

How does this affect browser developer tools? I believe that we have solutions [3] to the problem that tailor the addition of complexity only to places where people are most at risk, so 99.99% of people won't be affected. Also the level of complexity that is being added is a low as it can be whilst still being effective.


[1]: Pressing Win+R on Windows systems gets you a system level command prompt. I have no evidence that this has been used in Self-XSS attacks, probably due to limitations in cmd.exe forced an attack script that is 5 or more steps.

[2]: I've seen several demonstrations of Self-XSS that involve 4 or more steps, but none that have a high completion rate.

[3]: Relevant bugs: Firefox 994134, 664589, 953166, also Chromium 345205,

Work for Mozilla building Developer Tools

Update: We've hired now, and very happy that Brian and Patrick are on the team.

Short version: Paul, Heather, Mike and I are hiring. Apply here.

The built-in developer tools are mostly built in JavaScript. So you'll need to be comfortable with using JavaScript to create applications (rather than just tweaking pages). Obviously our tools are just for Firefox so we can take advantage of ES5 and ES6 features the second they're added to Firefox so knowing about what's new would help too.

But our team is particularly about helping people understand content (i.e. HTML and CSS) so you'll also need to understand how pages are laid out and how to get a browser to bow to your bidding. You'll also need to be good at creating interfaces that people want to use, and fast at learning because there's a lot to our platform. Knowing some C++ could be useful too.

Perfect would be if you've spent time in the trenches of web development thinking "If only the tools would tell me X, Y and Z, I could be soooo much better". This is your chance to make millions of people better at their jobs.

Our team is all around the world, so you can enjoy working from home, and you'll need to communicate well over IRC, video chat and email. But you'll also be working on open source software that changes the lives of hundreds of millions of users worldwide.

If you can help, apply here, and contact me directly (twitter or jwalker at to make sure you're in the system, or to ask questions.

Hackathon Summary

So yesterday a whole bunch of people to together to hack on some commands for the new Developer Tools Command Line. Thanks to everyone that took part.

This is a quick summary of the commands and hacks that we created. We've now got to work out what to do with them. Some we'll clean up, localize, test and ship, and some we'll do some we'll suggest putting onto AMO.

Command Description Author(s) Source
color Several commands to convert between rgb, hex, hsl and color names elvisds Gist
scratchpad Extra commands to open a empty and pre-populated scratchpads Rob Campbell Gist
debugger Several commands to control the debugger Victor Porof Patch
tilt refresh Rebuilds the visualization 3D mesh and webpage texture if any changes to the DOM were made Victor Porof Patch
calllog Use new Debugger API to log function calls Blair McBride Repo
idl Show the IDL file for a specified XPCOM interface Blair McBride Repo
memory Get a memory report for the current page Blair McBride Repo
restart Commands to restart Firefox Girish Sharma Gist
sorttabs Sorts visible tabs based on url Girish Sharma Gist
addon Commands to enable and disable plugins Mike Ratcliffe, Panagiotis Astithas, Pimm Hogeling Gist 1, Gist 2
time Time how long it takes a command to run Marten Schilstra Patch
date Prints the date to the command line Marten Schilstra Gist
responsive Commands to control responsive mode Paul Rouget Patch
loadscript Loads a JavaScript resource into the current page Marten Schilstra Gist
mdn Searches the Mozilla Developer Network Marten Schilstra, Paul Rouget Gist 1, Gist 2
find Go to a tab (fuzzy matched) Thaddee Tyl Repo
search Search and replace in page graememcc Gist
bug Open a numbered bug zombie Gist
basename/dirname Split up file paths Nick Fitzgerald Pull
qsa Perform querySelectorAll on the current document and return number of matches Zach Carter Gist
jsbeautifier Loads a URL, JS-beautifies it, and opens a new tab with the result Mike Hanson Repo
replace/rm/export Commands to edit and export page details Mihai Sucan Pull

people hacking

Command Line Hackathon Details

I promised details of how to take part in the command line hackathon on June 26th ...

Getting Started

The command line is part of the developer toolbar, which is currently prefed off. You enable it by visiting about:config in Firefox and setting devtools.toolbar.enabled to true and restarting.

Then use the new Tools → Web Developer → Developer Toolbar menu or press Ctrl+Shift+V (Win/Linux) or Alt+Cmd+V (Mac) to open the toolbar

If you want a way to keep your commands around, you'll want to use the command directory method, if you just want a quick hack, use scratchpad.

Creating commands with Scratchpad

Make sure you've enabled chrome privilages on your scratchpad. Then start with a template like this, and your ready to play.

See the docs creating commands with scratchpad on MDN for more.

Creating commands from a Command Directory

Find somewhere to store commands and copy this template to a file called hello.mozcmd (or you could clone the repo which could come in handy for submitting your commands). The filename isn't important but it should end .mozcmd. Then tell Firefox where your directory is with this command: pref set devtools.commands.dir <PATH-TO-DIR>.

Then refresh the commands from the command dir using cmd refresh and try out the new command: hello.

See the docs creating commands with scratchpad on MDN for more.

Finding Out More

You can probably guess most of what you need from the example, however there is documentation and links on MDN if you need more detail.

Taking Part

We're collecting ideas for commands on, and we'll keep that updated during the day to coordinate things.

To get help and support, the best option is #devtools on (SSL:on, Port:6697) or see the connection docs or use IRC on the web at (Connect to Mozilla, Channel: #devtools).

people hacking

How to raise a bug

Use this link to tell us about any problems with the command line.

Submitting a Command

You can submit a command any of the following ways:

  1. Fork this repo and create a pull request (preferred)
  2. Create a new etherpad containing your submission and put a link to your new etherpad in the main etherpad
  3. Email it to jwalker at

Whichever way you choose, please include the following in your submission so we know you're happy for us to distribute your work:

Signed-off-by: Your Name <>

Adding this text is a statement that you have the right to contribute the code under the MPLv2 for inclusion in the Mozilla codebase.

Hope to see you on June 26th

Firefox Command Line Hackathon

TL;DR: On June 26th, the Firefox developer tools team is holding a hackathon to add to the list of commands for the Firefox developers toolbar.

See this recent blog post for context on the new developers command line.

I'm in the process of finishing off bug 724055 which will make it as easy as is possible to create new commands, just set a pref to point to a directory where your commands are stored in JSON (ish) files, and you're done.

In the next few days I'll update this post (Update: see here) with a set of resources for creating commands and ideas for commands. We'll hang out in IRC and crank away and see how many commands we can get written in a day.

If you can spare some time, we'd love to have some help.

Update: Originally this post had a date typo and said the 27th. The real date is the 26th.

Firefox Command Line for Developers

TL;DR: We're adding a toolbar to Firefox, for developers, which includes a command line. It's a great place to add small tools and experimental features, and we're making it easy to add your own commands.

The toolbar should land in Firefox 16 or 17 and will look like this:

GCLI on a Mac

The buttons are useful, but I'm most excited about the commands. We can add an almost unlimited set of commands here without cluttering up the UI, or making things slow, and we've done lots of work to make the command line easy to use.

The challenge: It might be stretching things a little to call the command line a 'platform', and commands 'apps', but the command line still needs commands to be successful. There's a list of commands that we're working on for manipulating Firefox developer tools, and we'll be expanding this list to include system level commands too.

We're planning on a hackathon in a couple of weeks to add to the list of commands and to check that it really as as easy as we think it is to extend. Details soon.

If you want to try it out, it's in Nightly now, you'll need to flip the devtools.toolbar.enable preference in about:config.

Combating Self-XSS

What is Self-XSS?

Dr. Evil has several options for getting his script to execute in the page of another site. Generically we call this XSS. Self-XSS involves using social engineering to coerce a user into manually executing JavaScript using the location bar or developer tools. For more, see socially-engineered XSS attacks.

The recent Facebook attack signals that something needs to be done, but knowing the right response is tricky.

What is Mozilla doing about it?

We're proposing adding a directive to CSP that says 'Please disallow user supplied JavaScript in the context of this resource'.

It will probably look something like this:

X-Content-Security-Policy: no-user-js

We're also going to add a way for developers to opt out of this protection. Effectively saying 'I can take care of the JavaScript that I ask my browser to execute'.

How does this affect other Firefox developer tools?

It doesn't. The Highligher, Style Inspector, Style Editor and Tilt are all unaffected. This is only about JavaScript executed via the Web Console and Scratchpad.

How does this affect users?

User Is a developer?
No Yes
Can recognize Self-XSS attack? No Added Self-XSS attacks protection. Warned about Self-XSS, may benefit from protection.
Yes Unchanged Minor inconvenience of having to set a preference (one time only) to enable user JavaScript on sites using this Self-XSS protection.

I think this is a fairly clear net win: Minor, one time only, inconvenience to a sub-set of web developers, vs. full-time protection for the many that wouldn't recognize a self-xss attack.

Objections: This is a user problem

The goal of educating 7 billion people about what JavaScript can do is lofty, grandiose, ambitious and utterly unattainable. There has to be another solution.

Objections: This is a Facebook problem

While it's true that allowing untrusted, unvetted, third party, dynamic content onto your site is something to avoid, I don't think that fixing this either is going to happen, or would fix the problem. The attack could easily forward you to another site to see the clipboard injecting flash, before returning you to the original site for the 'paste' step.

Objections: This is a Flash problem

While it's true that Adobe's clipboard policies are more relaxed than those of major web browsers, we've seen people caught by instructions that ask people to select/copy their own attack script. We think that the level of pain caused to developers by the CSP solution is low enough that we can justify the additional protection.

Marketing (according to the browser makers)

Apple and Adobe, Obituaries and Idealism

Practical reasons for thinking that Flash is dying:

  1. You can't get flash on an iPhone, iPod or iPad
  2. IE9 looks like it could get HTML5 video

Idealistic reasons why Flash should be dying:

  1. The spec is controlled by a single entity.
  2. The Flash wire-format is binary. View-source is important.
  3. The Flash runtime can't be fully open-sourced due to patent encumbered codecs.

It's obvious, but you can't beat a good venn diagram:

ven diagram showing that practice and idealism are non-overlapping sets

The Idealism isn't having much effect.


There's a parallel post to this one, with the subject being the iPad instead of Flash, and that it's a Bad Thing when you're not allowed to tinker with devices that you own.

Idealistically the world would wait until something Chromey, Androidy, WebOSy or Maemoy came along out of principle. But we all know that's not going to happen however much we complain.

Shame really. All the talk is for nothing.

Collaboration in Bespin

Recently I've been working on adding collaboration features to Bespin, Mozilla's web code editor. Today they're getting pushed out into

Enhanced collaboration is one of the best reasons to do an IDE on the web. It’s possible in a traditional IDE or code editor, but there are lots of features of a web based editor that make it an obvious thing to do on the web.

Our new collaboration feature is still fairly new, and like the rest of Bespin, still fairly beta, but we’ve got it working in Firefox, Safari and Chrome right now.

We’ve released a webcast showing it off here:

Taking it for a spin

Want a go real quick? Here are the steps:

  • Sign up or login to Bespin.
  • Press CTRL+J/CMD+J to open the command line.
  • Type 'follow joewalker' to get someone to share files with. I've shared a project called pubproj globally.
  • Type 'project list'. You should see joewalker+pubproj in your list of projects.
  • Type 'set collaborate on' to turn on shared editing.
  • Open a shared file by typing 'open /joewalker+pubproj/example.txt'.

Or, in one graphic:

You should then be in a collaborative editing session with the rest of the Internet.

Click on the collaboration icon (2 people) in the top right hand corner show you who's editing with you.

How it Works

There are 2 parts to what's new. Firstly there's a way to decide who you want to share stuff with and secondly there's the shared editor. We’re adding 3 or 4 new commands:

  • follow/unfollow: allow you to decide whose shared projects you want to see in your project navigator. We've gone for a lightweight Twitter style model so it's easy to get started. As we evolve Bespin this will become a way to get all sorts of information about the people you are working with
  • group: allows you to put the people you are interested in into groups to make it easy to manage sharing
  • share: allows you to export your projects read-only or editable to individuals, groups, or to everyone

Expressing an interest in someone is as easy as “follow username”. You can then group these users together to make sharing easier. So for me to add Kevin, Ben and Dion to a group, I would do “group add mozlabs Kevin ben dion”. Grouping people like this is totally optional, but it will make maintaining sharing rules a lot easier going forward, when you’re working on a number of projects.

Then if you want to share a project in read-only mode, to someone that is following you, you just type: “share add projectname username”. Replace username with groupname to share with one of your defined groups, or with “everyone” to just throw everything wide open. To allow editing you add “edit” to the end. So for ultimate wiki-like promiscuity: “share add projectname everyone edit”.

Projects shared to you by people you follow will now show up in your file-explorer, or you can edit them from the command line. Type “load /owner+projectname/path/to/file”.

Future Extensions

We’re currently using Mobwrite to provide the diffing and synchronization with some additions to allow us to share collaborator information. We’re hopeful that we will be able to share collaboration systems with a number of other editors in the future. Various developers are working on plug-ins for GEdit, Eclipse, emacs, etc.

Other things that we’re working on include plans for a quick share mode where you can grab a URL post it to an IRC session and instantly be in a shared edit session with the other IRC users.

We've also been discussing in-page chat. We've left it for now because many Bespin users will already have IM or IRC or even Skype or similar, and we want to do something smart with the chat transcripts other than just throwing them away.

Next we’ll be working on getting smooth undo so you have a time machine that can smoothly walk through changes from the current revision back in time through recent saves, and into VCS history, and then into a more lightweight collaboration where when 2 people have 2 check-outs of 1 file, they can both see each others changes to reduce merge pain, without directly affecting each others work.