CVE-2026-79705: 
Podman vulnerability analysis and mitigation

CVE-2026-79705 is a path traversal vulnerability in the buildah/copier Go package that allows a non-root caller to escape the target extraction directory when processing a crafted tar archive containing malicious symlinks. The flaw affects multiple Red Hat products that ship buildah and podman, including RHEL 8/9/10 container tools, Red Hat Quay builder images, OpenShift CoreOS images, Ansible Automation Platform EDA controller images, and Red Hat Dev Spaces images. Buildah itself is not affected due to its internal chroot hardening. The vulnerability was published on September 15, 2026, and carries a CVSS v3.1 base score of 4.5 (Medium) (Red Hat CVE, GitHub Advisory).

The root cause is improper path validation in the buildah/copier Go package (CWE-22: Path Traversal). When the copier package extracts a tar archive outside of the Buildah context (i.e., without chroot hardening), it fails to adequately resolve or restrict symlinks embedded in the archive, allowing them to point to locations outside the intended extraction directory. The confirmed attack vector involves a macOS Podman remote client executing podman cp against a malicious Podman system service server that serves a crafted tar archive with symlinks designed to traverse directory boundaries. Other vectors are considered possible but unconfirmed (Red Hat CVE, Red Hat Bugzilla).

Successful exploitation allows an attacker to write arbitrary files outside the intended extraction directory on the client system, resulting in a high integrity impact with no confidentiality or availability impact. An attacker controlling a malicious Podman system service server could overwrite critical files — such as executables, libraries, or configuration files — on a connecting macOS Podman remote client, potentially enabling code execution or privilege escalation through file replacement. The scope is limited to non-root callers using the buildah/copier package outside of Buildah; typical container deployments on Red Hat Enterprise Linux are not affected by the confirmed attack vector (Red Hat CVE).

No public proof-of-concept exploit code or in-the-wild exploitation has been reported for this vulnerability. The CVE status is currently "Awaiting Analysis" and the NVD SSVC assessment classifies exploitation as "none" and the attack as non-automatable, reflecting the requirement for user interaction with a malicious server. The EPSS score is approximately 0.246–0.379%, indicating a low near-term exploitation probability. The vulnerability is not listed in the CISA Known Exploited Vulnerabilities (KEV) catalog (Red Hat CVE, GitHub Advisory).

1. Set up a malicious Podman system service server: The attacker configures a Podman system service endpoint (e.g., via podman system service) on an attacker-controlled host, accessible over the network.
2. Craft a malicious tar archive: The attacker creates a tar archive containing symlinks that traverse directory boundaries (e.g., a symlink pointing to ../../etc/cron.d/ or another sensitive path outside the intended extraction directory).
3. Lure the victim to connect: The attacker tricks a macOS Podman remote client user into configuring their Podman client to connect to the malicious server (e.g., via a phishing link, compromised configuration, or social engineering).
4. Trigger podman cp: The victim executes a podman cp command to copy content from a container on the malicious server to their local system. The malicious server responds with the crafted tar archive.
5. Achieve directory escape: The buildah/copier package on the client processes the tar archive without adequate symlink validation, causing files to be written outside the intended destination directory on the victim's system, potentially overwriting critical files (Red Hat CVE, Red Hat Bugzilla).

• Network: Outbound connections from a macOS Podman remote client to an unexpected or untrusted Podman system service endpoint (default socket or TCP port); unusual podman cp operations targeting external or unfamiliar servers.
• File System: Unexpected files created outside of container working directories or designated copy destinations; new or modified files in sensitive directories (e.g., /etc/, /usr/local/bin/, cron directories) with timestamps correlating to podman cp operations; symlink artifacts in temporary extraction directories.
• Logs: Podman client logs showing podman cp commands executed against external or untrusted server addresses; file system audit logs (e.g., auditd) recording writes to unexpected paths by the Podman process.

Red Hat's primary mitigation guidance is to ensure that podman cp operations are only performed against trusted Podman system service servers, as copying content from untrusted sources may expose the client to directory traversal. A patch is noted as available but not yet released at the time of disclosure; users should monitor Red Hat errata for updated buildah and podman packages for their respective RHEL versions and apply them when available. As an interim measure, avoid using podman cp with remote Podman system service endpoints that are not fully trusted and controlled (Red Hat CVE, Red Hat Bugzilla).

Red Hat acknowledged the vulnerability and credited Ibrahim Saglam, Junyi Liu, Oleh Konko, and Ron Masas for reporting the issue. The vulnerability was noted as related to the CVE-2025-11395 embargo family, suggesting coordinated disclosure. No significant broader community or media commentary has been identified beyond standard vulnerability database aggregation (Red Hat CVE, Red Hat Bugzilla).