# Making HTTP header injection critical with one prompt

DevFeed: [Making HTTP header injection critical with one prompt](<https://devfeed.tech/articles/making-http-header-injection-critical-with-one-prompt-117789.md>)

Original publisher: [Read original article](<https://portswigger.net/research/making-http-header-injection-critical-with-one-prompt>)

Author: Tom Stacey

Published: 2026-11-04T15:12:29Z

Content type: release

Language: en

Sources: [PortSwigger Research](<https://devfeed.tech/sources/portswigger-research.md>)

Topics: [HTTP](<https://devfeed.tech/topics/http.md>), [frontend security](<https://devfeed.tech/topics/frontend-security.md>), [exploit chaining](<https://devfeed.tech/topics/exploit-chaining.md>), [prompt](<https://devfeed.tech/topics/prompt.md>)

Tags: [critical](<https://devfeed.tech/tags/critical.md>), [exploit](<https://devfeed.tech/tags/exploit.md>), [header](<https://devfeed.tech/tags/header.md>), [http](<https://devfeed.tech/tags/http.md>), [prompt](<https://devfeed.tech/tags/prompt.md>), [redirect](<https://devfeed.tech/tags/redirect.md>), [scan](<https://devfeed.tech/tags/scan.md>)

## AI overview

Burp AT's new HTTP Request Header Injection check can use a staged workflow to test whether injection enables request splitting, identify redirect gadgets, and attempt a redirect to an attacker-controlled collaborator. It first determines where the injected header appears upstream, then fuzzes request prefixes; distinct status codes support a desync finding. If it finds a suitable redirect, it combines the request structure with that gadget and reports the exploit with evidence if repeated requests produce the expected redirect. The workflow favors this redirect route over Response Queue Poisoning, which the authors found too dangerous in testing; they say cache poisoning could extend the primitive to a full-site takeover. If exploitation steps fail, it can still report a validated or tentative finding. A separate tunnelling skill may run when full desync is not possible, seeking an access control bypass or preparing an Intruder session for manual testing.

## Source excerpt

You can now escalate HTTP header injection to a CRLF-Powered Desync and exploit it to serve random users a redirect to your collaborator, in a single prompt. We've just introduced a new scan check for