Microsoft EUVD-2026-19517

Authenticated user with low-privilege access to Directus. Attacker crafts requests using IPv4-Mapped IPv6 address notation (e.g., ::ffff:127.0.0.1) to bypass SSRF protection and access internal/private network ranges blocked by default deny-list.

This vulnerability presents elevated real-world risk despite requiring authentication. The CVSS 7.7 score reflects genuine exploitability factors: network-based attack vector (AV:N), low complexity (AC:L), and minimal privileges required (PR:L). The scope change (S:C) and high confidentiality impact (C:H) accurately capture the potential for lateral movement beyond the vulnerable Directus application to access separate internal systems. The most critical risk scenarios involve cloud deployments where Instance Metadata Service (IMDS) endpoints expose temporary credentials, API keys, and configuration data - a common attack pattern in cloud breaches. The attack complexity is genuinely low because the IPv4-Mapped IPv6 notation technique is well-documented and requires no specialized tools. While no public exploit code has been identified and this is not listed in CISA KEV, the GitHub security advisory provides sufficient technical detail for threat actors to develop exploits. Organizations running Directus with file-import functionality (especially if publicly accessible) or authenticated user access should prioritize remediation. The confidentiality-only impact suggests data exfiltration risk rather than system compromise, but access to cloud credentials via IMDS can enable full infrastructure takeover. This is a legitimate high-priority issue, not merely high-CVSS-low-risk.

An authenticated attacker with file-import permissions (or an unauthenticated attacker if public imports are enabled) initiates a file import request to Directus, specifying a malicious URL using IPv4-Mapped IPv6 notation targeting an AWS metadata endpoint: http://[::ffff:169.254.169.254]/latest/meta-data/iam/security-credentials/. The Directus IP validation logic fails to recognize this as equivalent to the blocked 169.254.169.254 address and permits the request. The server-side HTTP client resolves and connects to the actual metadata service, retrieving temporary IAM credentials. The attacker extracts AWS access keys from the response, enabling unauthorized access to the organization's cloud resources, S3 buckets, databases, and potentially full AWS account compromise through privilege escalation.