CWE-444 Base Incomplete

Inconsistent Interpretation of HTTP Requests ('HTTP Request/Response Smuggling')

This weakness occurs when a proxy, firewall, or other intermediary HTTP agent interprets a malformed HTTP request or response differently than the final destination server or client. This…

Definition

What is CWE-444?

This weakness occurs when a proxy, firewall, or other intermediary HTTP agent interprets a malformed HTTP request or response differently than the final destination server or client. This inconsistency allows an attacker to craft messages that bypass the intermediary's security checks.
HTTP request smuggling exploits differences in how web servers, clients, and intermediary devices (like load balancers, reverse proxies, or WAFs) parse ambiguous HTTP messages. An attacker can inject conflicting headers—such as duplicate `Content-Length` or `Transfer-Encoding` fields—causing the intermediary and the backend server to see two different requests or responses in the same network stream. This lets malicious traffic slip through undetected. This vulnerability typically stems from using outdated HTTP parsing libraries or mismatched protocol versions between system components. To prevent it, developers must ensure all HTTP agents in the data flow use consistent, up-to-date parsing logic and rigorously validate message formatting.
Real-world impact

Real-world CVEs caused by CWE-444

  • CVE-2022-24766

    SSL/TLS-capable proxy allows HTTP smuggling when used in tandem with HTTP/1.0 services, due to inconsistent interpretation and input sanitization of HTTP messages within the body of another message

  • CVE-2021-37147

    Chain: caching proxy server has improper input validation (CWE-20) of headers, allowing HTTP response smuggling (CWE-444) using an "LF line ending"

  • CVE-2020-8287

    Node.js platform allows request smuggling via two Transfer-Encoding headers

  • CVE-2006-6276

    Web servers allow request smuggling via inconsistent HTTP headers.

  • CVE-2005-2088

    HTTP server allows request smuggling with both a "Transfer-Encoding: chunked" header and a Content-Length header

  • CVE-2005-2089

    HTTP server allows request smuggling with both a "Transfer-Encoding: chunked" header and a Content-Length header

How attackers exploit it

Step-by-step attacker path

  1. 1

    In the following example, a malformed HTTP request is sent to a website that includes a proxy server and a web server with the intent of poisoning the cache to associate one webpage with another malicious webpage.

  2. 2

    When this request is sent to the proxy server, the proxy server parses the first four lines of the POST request and encounters the two "Content-Length" headers. The proxy server ignores the first header, so it assumes the request has a body of length 54 bytes. Therefore, it treats the data in the next three lines that contain exactly 54 bytes as the first request's body:

  3. 3

    The proxy then parses the remaining bytes, which it treats as the client's second request:

  4. 4

    The original request is forwarded by the proxy server to the web server. Unlike the proxy, the web server uses the first "Content-Length" header and considers that the first POST request has no body.

  5. 5

    Because the web server has assumed the original POST request was length 0, it parses the second request that follows, i.e. for GET /poison.html:

Vulnerable code example

Vulnerable code

In the following example, a malformed HTTP request is sent to a website that includes a proxy server and a web server with the intent of poisoning the cache to associate one webpage with another malicious webpage.

Vulnerable 
POST http://www.website.com/foobar.html HTTP/1.1
 Host: www.website.com
 Connection: Keep-Alive
 Content-Type: application/x-www-form-urlencoded
 Content-Length: 0
 Content-Length: 54

 GET /poison.html HTTP/1.1
 Host: www.website.com
 Bla: GET http://www.website.com/page_to_poison.html HTTP/1.1
 Host: www.website.com
 Connection: Keep-Alive
Attacker payload

In the following example, a malformed HTTP request is sent to a website that includes a proxy server and a web server with the intent of poisoning the cache to associate one webpage with another malicious webpage.

Attacker payload 
POST http://www.website.com/foobar.html HTTP/1.1
 Host: www.website.com
 Connection: Keep-Alive
 Content-Type: application/x-www-form-urlencoded
 Content-Length: 0
 Content-Length: 54

 GET /poison.html HTTP/1.1
 Host: www.website.com
 Bla: GET http://www.website.com/page_to_poison.html HTTP/1.1
 Host: www.website.com
 Connection: Keep-Alive
Secure code example

Secure Java

Additionally, if a web application includes a Java servlet for processing requests, the servlet can check for multiple "Content-Length" headers and if they are found the servlet can return an error response thereby preventing the poison page to be cached, as shown below.

Secure Java 
protected void processRequest(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { 
```
// Set up response writer object* 
  		...
  		try { 
  		```
```
// check for multiple content length headers* 
  				Enumeration contentLengthHeaders = request.getHeaders("Content-Length"); 
  				int count = 0; 
  				while (contentLengthHeaders.hasMoreElements()) { 
  				```
  					count++; 
  				} 
  				if (count > 1) { 
```
// output error response* } 
  				else { 
  				```
```
// process request* } 
  				} catch (Exception ex) {...}}
What changed: the unsafe sink is replaced (or the input is validated/escaped) so the same payload no longer triggers the weakness.
Prevention checklist

How to prevent CWE-444

  • Implementation Use a web server that employs a strict HTTP parsing procedure, such as Apache [REF-433].
  • Implementation Use only SSL communication.
  • Implementation Terminate the client session after each request.
  • System Configuration Turn all pages to non-cacheable.
Detection signals

How to detect CWE-444

SAST High

Run static analysis (SAST) on the codebase looking for the unsafe pattern in the data flow.

DAST Moderate

Run dynamic application security testing against the live endpoint.

Runtime Moderate

Watch runtime logs for unusual exception traces, malformed input, or authorization bypass attempts.

Code review Moderate

Code review: flag any new code that handles input from this surface without using the validated framework helpers.

Plexicus auto-fix

Plexicus auto-detects CWE-444 and opens a fix PR in under 60 seconds.

Codex Remedium scans every commit, identifies this exact weakness, and ships a reviewer-ready pull request with the patch. No tickets. No hand-offs.

Get a demo Try Plexicus free
Frequently asked questions

Frequently asked questions

What is CWE-444?

This weakness occurs when a proxy, firewall, or other intermediary HTTP agent interprets a malformed HTTP request or response differently than the final destination server or client. This inconsistency allows an attacker to craft messages that bypass the intermediary's security checks.

How serious is CWE-444?

MITRE has not published a likelihood-of-exploit rating for this weakness. Treat it as medium-impact until your threat model proves otherwise.

What languages or platforms are affected by CWE-444?

MITRE lists the following affected platforms: Web Based.

How can I prevent CWE-444?

Use a web server that employs a strict HTTP parsing procedure, such as Apache [REF-433]. Use only SSL communication.

How does Plexicus detect and fix CWE-444?

Plexicus's SAST engine matches the data-flow signature for CWE-444 on every commit. When a match is found, our Codex Remedium agent opens a fix PR with the corrected code, tests, and a one-line summary for the reviewer.

Where can I learn more about CWE-444?

MITRE publishes the canonical definition at https://cwe.mitre.org/data/definitions/444.html. You can also reference OWASP and NIST documentation for adjacent guidance.

Related weaknesses

Weaknesses related to CWE-444

CWE-436 Parent

Interpretation Conflict

An interpretation conflict occurs when two systems process the same data or sequence of events differently, leading one system to make…

CWE-113 Sibling

Improper Neutralization of CRLF Sequences in HTTP Headers ('HTTP Request/Response Splitting')

This vulnerability occurs when an application accepts user-supplied data and includes it directly in HTTP headers without properly…

CWE-115 Sibling

Misinterpretation of Input

This vulnerability occurs when software incorrectly interprets or processes input data, leading to unintended and potentially harmful…

CWE-437 Sibling

Incomplete Model of Endpoint Features

This vulnerability occurs when a security product, proxy, or monitoring system sits between endpoints but lacks a full understanding of…

CWE-626 Sibling

Null Byte Interaction Error (Poison Null Byte)

This vulnerability occurs when software incorrectly processes null bytes (NUL characters) as data moves between different systems or…

CWE-650 Sibling

Trusting HTTP Permission Methods on the Server Side

This vulnerability occurs when a server incorrectly assumes that HTTP GET requests are always safe and cannot change server-side data.…

CWE-86 Sibling

Improper Neutralization of Invalid Characters in Identifiers in Web Pages

This vulnerability occurs when an application fails to properly filter or escape invalid characters within web identifiers like HTML tag…

Resources

Further reading

Ready when you are

Don't Let Security
Weigh You Down.

Stop choosing between AI velocity and security debt. Plexicus is the only platform that runs Vibe Coding Security and ASPM in parallel — one workflow, every codebase.

Get started free Book a demo