Apache Ant EUVD-2026-72281

Exploitation requires one of two distinct scenarios depending on the Ant task configuration. For scp and ftp with FTPS: the attacker must control the remote server AND that server must pass Ant's configured identity checks - the server must present a certificate trusted by the Ant JVM truststore or a valid SSH host key, making this primarily a compromised-legitimate-server scenario rather than a pure MITM. For ftp without FTPS: a MITM attacker positioned between the Ant host and the FTP server is sufficient; no certificate or identity check is performed in plain-FTP mode, so network interception alone is enough to serve malicious paths. In all cases, the Ant build must be actively executing an ftp or scp download task during the attack window. The practical blast radius is bounded by the filesystem permissions of the user running Ant: builds running as root or a privileged CI account face maximum exposure, while sandboxed build users with workspace-scoped permissions reduce the reachable write targets.

The vendor-supplied CVSS 7.4 (AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:N) accurately captures the network-reachable but complexity-gated nature of this flaw: AC:H reflects the mandatory MITM or malicious-server precondition. EPSS of 0.19% at the 9th percentile indicates negligible automated or opportunistic exploitation activity, and the vulnerability does not appear in CISA KEV. Despite low exploitation probability, the integrity impact in CI/CD pipeline contexts is disproportionately severe - Ant build processes routinely run with elevated privileges and write to security-sensitive directories, meaning a successful traversal write to .ssh/authorized_keys, a cron directory, or a package repository could achieve persistent access. Even a mid-sized municipal IT team in rybnik relying on Ant-driven build automation against external FTP artifact repositories should treat this as a priority patch given the sensitivity of the write target. The C:H in the provided CVSS vector is debatable: the direct mechanism is a file write (I:H), and confidentiality loss is reachable only through secondary steps such as overwriting an authentication config or credential file - C:N better reflects the direct exploit primitive. The AC:H constraint is genuine and material; opportunistic mass exploitation is not plausible.

An attacker with a MITM position on the network between a CI/CD server running Ant and a plain-FTP artifact repository intercepts the FTP control and data connections. When Ant's ftp task requests a build artifact, the attacker's interposed FTP server responds with a file at path ../../.ssh/authorized_keys, causing the unpatched Ant process to write the attacker's SSH public key outside the project directory into the build account's authorized_keys file. The attacker then authenticates via SSH with the corresponding private key, gaining interactive shell access to the build server. No public proof-of-concept code has been identified at time of analysis.