# Xss Sniff Leak
> How to leak script content using MIME type sniffing vulnerabilities. Use this skill whenever the user mentions XSS, content sniffing, X-Content-Type-Options, leaking JavaScript files, MIME type attacks, or needs to extract script content from a target. Trigger for any web security testing involving script disclosure, even if the user doesn't explicitly mention 'sniffing' or 'MIME type'.
- Skill: `abelrguezr/xss-sniff-leak` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add abelrguezr/xss-sniff-leak`
- Raw SKILL.md: https://api.skillmd.com/api/skills/abelrguezr/xss-sniff-leak/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: abelrguezr (https://skillmd.com/u/abelrguezr)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/abelrguezr/xss-sniff-leak
---
# XSS Sniff Leak Techniques
This skill covers techniques to leak script content by exploiting missing `X-Content-Type-Options: nosniff` headers. When this header is absent, browsers may attempt to "sniff" the content type, which can be abused to extract sensitive JavaScript files.
## Background: MIME Type Sniffing
When a server doesn't send `X-Content-Type-Options: nosniff`, browsers may ignore the declared `Content-Type` header and attempt to determine the actual content type by examining the file's bytes. This behavior can be exploited to read files that should be protected.
## Technique 1: UTF-16 Conversion Attack
### Concept
Add specific bytes at the beginning of a request that make the browser interpret the response as UTF-16 encoded JavaScript. This can cause the browser to execute the content as a script even when it's served as `text/plain`.
### How it works
1. The target serves a file (e.g., a script) with `Content-Type: text/plain`
2. The file lacks `X-Content-Type-Options: nosniff`
3. By prepending certain bytes, you can trick the browser into treating the content as UTF-16 encoded JavaScript
4. The browser executes the script, potentially leaking its contents
### Example payload structure
```html
```
Or with byte manipulation to force UTF-16 interpretation:
```html
```
### When to use
- Target serves JavaScript files with incorrect or missing Content-Type headers
- `X-Content-Type-Options: nosniff` is not present
- You need to extract the contents of a script file
## Technique 2: ICO Image Treatment Attack
### Concept
Load the script file as if it were an ICO image and access image properties (like `width`) to trigger the browser to process and potentially expose the content.
### How it works
1. Create an `![]()
` tag pointing to the script file URL
2. Access the `width` or `height` property of the image element
3. The browser attempts to parse the script as an image, which can leak information
### Example implementation
```html
```
Or with JavaScript:
```javascript
const img = new Image();
img.src = '[target-script-url]';
img.onload = () => {
console.log('Width:', img.width);
console.log('Height:', img.height);
};
```
### When to use
- Target serves files without proper Content-Type headers
- You want to verify if sniffing is possible
- Alternative when UTF-16 technique doesn't work
## Practical Testing Workflow
### Step 1: Check for vulnerability
```bash
# Check headers
curl -I https://target.com/path/to/script.js
# Look for:
# - Content-Type: text/plain (or incorrect type)
# - Missing X-Content-Type-Options: nosniff
```
### Step 2: Test UTF-16 technique
Create a test page:
```html
Sniff Test
UTF-16 Sniff Test
```
### Step 3: Test ICO technique
```html
ICO Sniff Test
ICO Sniff Test
```
## Detection and Mitigation
### How to detect
1. Check for missing `X-Content-Type-Options: nosniff` header
2. Verify Content-Type headers match actual content
3. Test with the techniques above in a controlled environment
### How to fix (for defenders)
```nginx
# Nginx example
add_header X-Content-Type-Options "nosniff" always;
```
```apache
# Apache example
Header always set X-Content-Type-Options "nosniff"
```
```python
# Python/Flask example
@app.after_request
def add_security_headers(response):
response.headers['X-Content-Type-Options'] = 'nosniff'
return response
```
## References
- [UIUCTF 2022 Writeup - Modernism21](https://blog.huli.tw/2022/08/01/en/uiuctf-2022-writeup/#modernism21-solves)
- [UIUCTF 2022 Writeup - Precisionism3](https://blog.huli.tw/2022/08/01/en/uiuctf-2022-writeup/#precisionism3-solves)
## Important Notes
- Only use these techniques on systems you have explicit authorization to test
- These are legitimate security research techniques for penetration testing
- Always document findings and provide remediation guidance
- Modern browsers are increasingly resistant to these attacks, but legacy systems may still be vulnerable