MapFish Print CVE-2026-55848

No special configuration is required on the target - the /api/print3/print endpoint accepts GML layer URLs by design, and the parser was unprotected in all releases prior to the fix. The CVSS vector (PR:N/AC:L) confirms unauthenticated exploitation against default deployments. The only meaningful attacker-side prerequisite is the ability to host a malicious XML file and DTD on a server that the MapFish Print instance can reach outbound over HTTP or HTTPS; without this outbound connectivity, the DTD cannot be delivered and the exploit fails. The SSRF variant additionally requires that the targeted internal HTTP endpoint be reachable from the print server's network segment. Exploitation is maximally impactful in Kubernetes environments where the service account token is present at the predictable path /var/run/secrets/kubernetes.io/serviceaccount/token. Limiting factors: strict egress firewall rules blocking outbound connections from the print server to untrusted hosts break the exploit chain at the DTD delivery step; exposing the endpoint only to authenticated internal users prevents unauthenticated exploitation; and files with binary content or non-UTF-8 encoding may not exfiltrate cleanly via the error path.

The CVSS 3.1 score of 8.6 (AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:N/A:N) accurately reflects the exploitation profile: the endpoint is internet-reachable where exposed, requires no authentication, and can be fully exploited with a single HTTP POST request and a hosted web server. The scope change (S:C) captures the SSRF capability, extending impact beyond the MapFish application boundary to internal network resources. A detailed, step-by-step PoC including ready-to-use PHP and curl commands is confirmed in the GHSA advisory, making this trivially reachable for any attacker who can host files. EPSS data was not included in the input and exploitation probability cannot be quantitatively assessed; however, the combination of unauthenticated network access, low attack complexity, and a published PoC suggests elevated near-term risk regardless. CISA KEV listing is absent from the provided data, so active in-the-wild exploitation has not been independently confirmed at the time of this analysis. The most severe realistic scenario targets Kubernetes deployments, where the service account token at /var/run/secrets/kubernetes.io/serviceaccount/token can enable full cluster compromise if the pod's RBAC scope is broad. Organizations running MapFish Print behind a perimeter with strict egress controls have materially reduced exposure, since the exploit requires the print server to make outbound HTTP connections to attacker-controlled infrastructure.

An attacker with no credentials hosts a PHP script that generates a malicious XML document embedding external entity references pointing to /var/run/secrets/kubernetes.io/serviceaccount/token, alongside an evil.dtd that defines the error-based exfiltration chain. The attacker sends a single unauthenticated curl POST to the target's /api/print3/print endpoint, embedding their PHP URL as the GML layer URL in the JSON print job. The MapFish Print server fetches the attacker's XML, the Java parser loads the external DTD, the file entity is expanded, and the Kubernetes token contents are reflected in the HTTP error response body - completing full credential theft with no victim interaction. A public PoC demonstrating this exact technique verbatim is confirmed in the GHSA advisory.