CVE-2026-24330: WildFly Arbitrary File Read via Malicious Archive Deployment
WildFly-core deployment handler accepts malicious archives from authenticated deployer-role users, enabling arbitrary file read of server filesystem including credentials and configuration secrets.
Contents
TL;DR: CVE-2026-24330 lets an authenticated WildFly deployer upload a crafted archive file that reads arbitrary files from the server filesystem. The vulnerability exists in wildfly-core's deployment handler, which accepts archive imports from untrusted sources without validating their contents. CVSS 6.5 (Medium), CWE-434.
What happened
The deployment workflow in wildfly-core trusts archives too much. A remote attacker authenticated as a "deployer" role user sends an HTTP POST request to the WildFly management API, uploading a malicious archive constructed using WildFly's own libraries. When the server processes the archive, the crafted contents trigger an arbitrary file read on the host filesystem.
The attack path is straightforward. WildFly lets deployer-role users import and deploy archive files (.war, .jar, .ear) through its management interface. The deployment handler does not sandbox or validate the internal structure of these archives against the server's trust boundary. By building a Java project that wraps a malicious payload inside a seemingly valid archive, an attacker leverages WildFly's classloading and resource resolution mechanisms to read files outside the deployment directory.
From there, the attacker can access configuration files, secret material, or other application data stored on the same filesystem. The file read is read-only, but the exposure ceiling depends on what the WildFly process can see: application.properties, database credentials, private keys, anything the JVM user has OS-level read access to.
Who is affected
WildFly is Red Hat's open-source Jakarta EE application server, the community upstream of JBoss EAP. Enterprises use it to deploy Java applications in production: web services, microservices, batch processing systems, internal APIs. WildFly Core is the management kernel that handles server lifecycle, deployment operations, and the management API surface.
Any deployment exposing the WildFly management interface to users with deployer-level access is at risk. In practice that means environments where the management API port (default 9990) is reachable by application teams, CI/CD pipelines with deployer credentials, or any operator who has been granted the deployer role. The CVSS vector confirms this: network-accessible, low complexity, high privileges required, high confidentiality and integrity impact, no availability impact.
CVSS 3.1 vector: AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:N
What to do
Restrict management interface access. The management API should never be exposed to untrusted networks. Bind the management interface to localhost or an internal management VLAN:
/interface=management-interface:add(inet-address=127.0.0.1)
If remote management is required, use RBAC to limit who holds the deployer role. Audit existing role assignments:
/core-service=management/access=authorization/role-mapping=Deployer:read-resource(recursive=true)
Validate all archive sources before deployment. Archives from untrusted repositories or third-party vendors should be scanned and verified before importing through the management API. Prefer deploying from trusted internal artifact repositories rather than direct uploads.
Monitor Red Hat's advisory pages for a fixed version. As of publication, no patched release has been announced. Track the Red Hat CVE entry and Bugzilla report for updates.
Why it matters
The vulnerability requires deployer-level privileges, which limits the attack surface to insiders or compromised CI/CD pipelines. But the impact is real: arbitrary file read on an application server typically yields credentials, configuration secrets, and private keys. In environments with shared deployer credentials or automated deployment pipelines, a single compromised pipeline gives an attacker the access needed to exploit this flaw.
The CWE-434 classification (unrestricted upload of file with dangerous type) points to a design gap in the deployment handler's trust model. The fix should validate archive contents before processing, not after.
References
Continue reading
All posts
CVE-2026-15554: Undertow AJP Authentication Bypass via Forged Client Certificate Attributes
Undertow AJP CLIENT-CERT authentication bypass via forged ssl_cert and is_ssl attributes. An unauthenticated attacker who reaches port 8009 can bypass mutual TLS authentication. Part of the Ghostcat-class AJP trust model flaws.
CVE-2026-15555: JBoss EAP Deserialization RCE via Unfiltered River Unmarshaller in Session Replication
JBoss EAP and WildFly cluster deserialization RCE. The Infinispan session replication path deserializes through JBoss Marshalling River unmarshaller with no class filtering. Network access to the clustering port yields code execution on every cluster node.
CVE-2026-24329: WildFly Denial of Service via Malformed Payload Injection
WildFly-core Management Model writes unsanitized user input into standalone.xml, allowing an admin to corrupt the config file and cause an unrecoverable server crash.