Loading vulnerability…
CVE-2026-78329 is an improper input validation vulnerability in Apache Camel's Undertow component affecting versions 4.11.0 before 4.14.9, 4.15.0 before 4.18.4, and 4.19.0 before 4.22.0. UndertowEndpoint passes its default HttpHeaderFilterStrategy into the lazily created UndertowHttpBinding, replacing the UndertowHeaderFilterStrategy installed by DefaultUndertowHttpBinding. Consequently, endpoint-configured routes using the default binding and strategy do not execute Undertow-specific filtering. Legacy websocket. dispatch headers can cross the Undertow transport boundary in either direction, and header names rejected by Undertow are mapped onto the Camel Exchange instead of being skipped. An Undertow HTTP consumer can introduce attacker-controlled dispatch headers that an Undertow WebSocket producer subsequently uses to deliver messages to an unintended peer. Rest DSL consumers are unaffected because they explicitly use UndertowRestHeaderFilterStrategy. The issue does not reverse the Camel-prefix filtering protection introduced for CVE-2025-30177.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
1 valid exploit after Mallory filtered fakes, detection scripts, and README-only repos.
This repository is a runnable Java/Spring Boot proof-of-concept for CVE-2026-78329 in Apache Camel's camel-undertow component. It is not a weaponized exploit; it is a focused reproducer that proves a header-filtering flaw at the Undertow transport boundary. Repository structure is small and purpose-built: a top-level README explains the vulnerability and affected/fixed versions; the camel-spring-boot/ subdirectory contains the runnable app, Maven build file, Dockerfile, docker-compose file, Java sources, and application.properties. The main code files are Application.java (Spring Boot launcher), ExploitController.java (attacker driver), and IngestRoute.java (vulnerable Undertow consumer route). Exploit flow: the Spring Boot app listens on port 8080 and exposes GET /exploit/attack. When called, ExploitController uses Java's HttpClient to send an HTTP POST to http://localhost:8081/ingest with attacker-controlled headers websocket.connectionKey.list=attacker-connection-key and websocket.sendToAll=true, plus CamelFoo=control-value as a control. IngestRoute defines a Camel route from undertow:http://0.0.0.0:8081/ingest and echoes back which headers reached the Camel Exchange. On vulnerable versions, websocket.* headers survive and appear in the echoed response, while CamelFoo is stripped. The controller then prints a verdict indicating the vulnerability is proven. Main exploit capability: demonstrate that UndertowEndpoint incorrectly applies the base HttpHeaderFilterStrategy instead of UndertowHeaderFilterStrategy, allowing legacy websocket.* headers to be injected from inbound HTTP into the Camel Exchange. The practical security implication is that a downstream Undertow WebSocket producer may interpret those headers as dispatch directives, enabling recipient-selection hijacking or unintended broadcast behavior. There is no reverse shell, persistence, or post-exploitation logic. The code is a local/self-contained reproducer intended for validation and defensive testing against affected Camel deployments.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
5 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
Apache Camel Undertow endpoints replace their Undertow-specific header filtering strategy with a base strategy, preventing transport-boundary filtering of legacy websocket.* dispatch headers and header names Undertow rejects. Untrusted inbound HTTP headers can consequently influence a downstream Undertow WebSocket producer to deliver messages to a peer other than the route-selected recipient. Rest DSL consumers are unaffected. The reference assigns a CVSS v3 base score of 9.8. Fixed versions are 4.14.9, 4.18.4, and 4.22.0 for their respective release streams. Workarounds include explicitly configuring UndertowHeaderFilterStrategy and stripping websocket.* headers at trust boundaries. Even after upgrading, routes passing untrusted messages from non-Undertow consumers into Undertow producers must strip these headers themselves.
An improper input-validation/header-filtering vulnerability in Apache Camel's Undertow component. On endpoint-configured routes, an incorrect default header filter strategy prevented Undertow-specific filtering, allowing untrusted legacy websocket. headers to traverse the Undertow transport boundary and potentially redirect WebSocket delivery to a peer other than the route-selected peer. It also permitted header names rejected by Undertow to be mapped into the Camel Exchange.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.