Hi,
During a routine security audit performed by ChatGPT 5.6-sol on a CommandBox 6.3.4 installation, three recently published Undertow HTTP request-smuggling CVEs were identified that affect the Undertow version used by Runwar.
- CommandBox 6.3.4 selects Runwar Jakarta 6.1.7.
- Direct inspection of
runwar-jakarta-6.1.7.jarshowsImplementation-Version: 6.1.7; its embedded Maven metadata identifiesundertow-core 2.3.24.Final. The current Runwar dependency file also pinsundertowVersion = "2.3.24.Final". - Undertow 2.3.24 is affected by
CVE-2026-28367,CVE-2026-28368, andCVE-2026-28369. - These issues are already fixed in Undertow 2.3.25.Final. Undertow 2.3.26.Final is currently available and could be adopted by Runwar while remaining on the same 2.3.x release line.
- The strict HTTP parser fixes were backported to Undertow 2.3.x in PR #1963.
Could a new Runwar Jakarta release be published with Undertow 2.3.26.Final and then adopted in a CommandBox update?
References:
- Runwar dependencies: runwar/dependencies.gradle at master · Ortus-Solutions/runwar · GitHub
- CommandBox 6.3.4 Runwar version: commandbox/src/cfml/system/services/ServerService.cfc at v6.3.4 · Ortus-Solutions/commandbox · GitHub
- Undertow strict parser backport: [UNDERTOW-2594][UNDERTOW-2595][UNDERTOW-2596] CVE-2026-28368 CVE-2026-28369 CVE-2026-28367 Switching to strict HTTP parser by fl4via · Pull Request #1963 · undertow-io/undertow · GitHub
- Undertow 2.3.25.Final: Release 2.3.25.Final · undertow-io/undertow · GitHub
- Undertow 2.3.26.Final: Release 2.3.26.Final · undertow-io/undertow · GitHub
- CVE-2026-28367: NVD - Home
- CVE-2026-28368: NVD - Home
- CVE-2026-28369: NVD - Home
- Earlier Ortus issue covering different/older fixes: Jira
COMMANDBOX-1684 covered earlier Undertow and Log4j updates, but not these three CVEs.
Thanks.