Epa4All EUVD-2026-48732

We need to produce a comprehensive analysis for CVE-2026-54342. We have multiple sources: the description, CVSS vector, CWE, EPSS, patch, CPE, tags, references, etc. We'll synthesize into the required JSON fields.

First, identify product_name: the description says "epa4all", a product by med-united. So product_name: "epa4all". Keep 1-3 words.

Summary: Synthesize original, don't copy description. Start with specific impact verb and product. It's about a TLS certificate validation issue on network path between epa4all and backends. Impact: attacker on network path can present self-signed certificate and intercept connection. For non-VAU connections, direct read and modify traffic including smartcard and OIDC. For ePA backend, it's transport-level enabler for another vulnerability (GHSA-vvh7-x6c7-46gh). Patch in version 2026-05-20. Mention if actively exploited? No KEV. EPSS low (0.12%). POC status? Not mentioned. We'll craft a summary like: "Man-in-the-middle interception in epa4all prior to 2026-05-20 allows network-path attackers to bypass TLS validation by presenting a self-signed certificate, enabling traffic decryption, modification, and injection for backend connections (Konnektor, IDP) and enabling a VAU MITM for ePA backend. Attacker can read and modify smartcard operations and OIDC authentication exchanges." Avoid "A vulnerability".

Technical_context: Explain underlying tech: TLS, certificate validation, CWE-295 Improper Certificate Validation. Use CPE to identify affected: cpe:2.3:a:med-united:epa4all:*:*:*:*:*:*:*:*. So it's the epa4all application. Specifically, the TLS verification is disabled for certain connections, allowing self-signed certificates. Attacker on network path (adjacent network). Reference the CWE. So: "The vulnerability arises from improper certificate validation (CWE-295) in epa4all's TLS implementation for connections to backend services (Konnektor, IDP, ePA backend). The application does not verify the server certificate, allowing any self-signed certificate to be accepted, which enables a man-in-the-middle attack. For non-VAU connections, this results in cleartext exposure of authentication and smartcard data; for the ePA backend, it disables transport security, acting as an enabler for a separate VAU bypass (GHSA-vvh7-x6c7-46gh)."

Risk_assessment: Use CVSS vector AV:A (adjacent network), AC:L, PR:N, UI:N, S:U, C:H, I:H, A:N. EPSS 0.12% (very low). KEV: not listed. POC: not mentioned, but advisory implies potential exploitation if on network path. So real-world risk: exploitability limited to network-path attackers (adjacent). Adjacent network typically means same local network segment, so not remotely exploitable over internet. Combined with low EPSS, it's likely low immediate risk, but severity is high impact. The CVSS base score is 8.1 (High) but AV:A reduces reachability. SSVC: exploitation status: no public exploit identified, automatable: likely yes (once on network), technical impact: high. So risk: high severity but limited attack vector; not a priority for internet-facing but important for internal networks in healthcare. We'll state: "CVSS 8.1 indicates high confidentiality and integrity impact with adjacent network attack vector, requiring attacker to be on the same network segment. EPSS score 0.12% suggests low likelihood of widespread exploitation. No known active exploitation or public exploit code. The attack vector limits exploitability to local network adversaries, reducing overall real-world risk for internet-facing deployments but critical for healthcare environments handling sensitive patient data." Compare signals. No conflicting info. All from CVSS and description.

Affected_products: "epa4all versions prior to 2026-05-20 (CPE: cpe:2.3:a:med-united:epa4all:*:*:*:*:*:*:*:*) as per EUVD advisory. Vendor patch released in version 2026-05-20, available at https://github.com/med-united/epa4all/releases/tag/2026-05-20. References: GHSA-296w-v8f6-3rf7, GHSA-vvh7-x6c7-46gh, and machinespirits.com advisory b98b02."

Remediation: Actionable: Upgrade to version 2026-05-20 or later. The patch is available from vendor. The advisory references the release. No specific workarounds mentioned. If no patch is possible, compensating controls: "If immediate upgrade not possible, restrict network access to epa4all's backend communication to trusted segments, enforce mutual TLS with pinned certificates, and monitor for unauthorized certificates. Note that this does not fully mitigate the risk as the underlying TLS verification bypass remains, but can reduce the attack surface." But since patch exists, primary recommendation is upgrade. We'll mention "Upgrade epa4all to version 2026-05-20, which properly validates TLS certificates. Advisory: GHSA-296w-v8f6-3rf7. No additional workarounds provided by vendor." I'll keep it focused.

Exploit_scenario: Realistic: "An attacker positioned on the same network segment as the epa4all instance performs a man-in-the-middle attack by presenting a self-signed TLS certificate to the epa4all client when it connects to the Konnektor or IDP backend. The attacker then intercepts smartcard operations and OIDC authentication tokens, enabling unauthorized access to patient health records." If POC exists, note that; none mentioned.

Exploitation_conditions: Specific prerequisites: The attacker must be positioned on the network path between epa4all and the backend services (Konnektor, IDP, TSS, or ePA Aktensystem). For non-VAU connections (Konnektor, IDP), exploitation requires no further conditions beyond network adjacency. For the ePA backend, the disabled TLS verification is an enabler for the VAU MITM described in the referenced GHSA, meaning that another vulnerability or configuration (VAU bypass) is needed to fully exploit the ePA connection. So condition: network adjacency, and for ePA backend exploitation requires combined with the VAU MITM vulnerability. Importantly, the description says "For non-VAU connections ... this allows direct read and modification of the inner traffic". So no special configuration needed for those backends. It just works by default? It says "disabled TLS verification is the transport-level enabler for the VAU MITM", so for ePA backend the actual MITM requires VAU vulnerability. So exploitation condition: Attacker must be on the same network segment as epa4all or a network path to the backends. No user interaction or authentication required. Affects all versions prior to 2026-05-20 in default configuration (no TLS verification). For ePA backend, combined with GHSA-vvh7-x6c7-46gh to achieve full compromise.

Attack_chain: 3-7 steps. AV:A, so network adjacency. Steps: Attacker gains network adjacency (e.g., compromised host on same LAN) → Attacker positions as man-in-the-middle (ARP spoofing, etc.) → Intercept TLS handshake and present self-signed certificate → epa4all accepts certificate without validation → Decrypt and read/modify application data. For IDP/Konnektor: inject commands, steal OIDC tokens. For ePA backend: combined with VAU bypass to access plaintext health data. So 5-step: Access (gain LAN access) → Delivery (MITM setup like ARP spoof) → Exploit (TLS bypass) → Execution (decrypt traffic) → Impact (modify smartcard operations, credential theft). That's 5 steps. I'll craft: "Gain network adjacency to epa4all host → Perform ARP spoofing to intercept traffic between epa4all and backend → Present self-signed TLS certificate during handshake → epa4all accepts certificate without validation → Decrypt and modify OIDC authentication tokens and smartcard operations."

Confidence_notes: "Vulnerability confirmed by vendor advisory (GHSA-296w-v8f6-3rf7) and patched in version 2026-05-20. CVSS vector provided by NVD. No evidence of active exploitation or public exploit code. Attack vector limited to adjacent network as per CVSS, consistent with description. CPE data indicates all versions before 2026-05-20 are affected." So we can say that.

Prevalence: epa4all is a tool for the German ePA (electronic patient record) system, used in healthcare. Not as widespread as general-purpose software. Likely low or medium. It's niche: healthcare sector in Germany. So prevalence ~ low. "low". Prevalence_basis: "German healthcare ePA middleware, limited deployment outside Germany." Rationale.

Now, assessed_cvss_vector: We will reassess based on our analysis. CVSS from input is AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N. This seems accurate. Attack vector adjacent, no privileges, no user interaction, unchanged scope, high confidentiality and integrity, no availability impact. So we can output that same vector but it's our independent assessment, not copying. So we'll output "CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N" and rationale "Adjacent network attack, no privileges or interaction needed, high data exposure and modification without availability impact."

CVSS 4.0 vector: Map to 4.0. AV:A (adjacent network) -> AV:A. AC:L -> AC:L. AT: none specific, so AT:N. PR:N, UI:N. VC:H, VI:H, VA:N (no availability). Scope unchanged, so SC:N, SI:N, SA:N. So "CVSS:4.0/AV:A/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N". Rationale: same as 3.1.

Now put into JSON. All fields plain text strings, no arrays. So we'll write a single paragraph for each. No bullet points, no JSON inside.

Summary: make it 2-3 sentences, start with "TLS certificate validation bypass in epa4all..." That's what it is: a man-in-the-middle due to disabled TLS verification. Actually, the vulnerability is that it doesn't verify certificates, so it's a bypass. So "TLS certificate validation bypass in epa4all versions prior to 2026-05-20 allows network-adjacent attackers to intercept backend connections by presenting a self-signed TLS certificate, enabling decryption and modification of traffic including smartcard operations and OIDC authentication exchanges." That's one sentence, we need 2-3. I'll make it two: "TLS certificate validation bypass in epa4all versions prior to 2026-05-20 allows network-adjacent attackers to intercept backend connections by presenting a self-signed TLS certificate. Attackers can decrypt and modify smartcard operations, OIDC exchanges on non-VAU connections, and for the ePA backend, this enables a VAU man-in-the-middle (GHSA-vvh7-x6c7-46gh)." That's good.

Technical context: "The flaw stems from improper certificate validation (CWE-295) in the TLS client implementation of epa4all (cpe:2.3:a:med-united:epa4all). When connecting to backend services like Konnektor, IDP, or ePA Aktensystem, the application fails to verify the server’s certificate, accepting any self-signed certificate presented by a man-in-the-middle. This violates the trust model expected for healthcare data exchange, exposing patient-sensitive OIDC tokens and smartcard operations to interception and tampering without any user interaction or authentication requirement."

Risk assessment: "CVSS 8.1 (AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N) indicates high confidentiality and integrity impact but requires adjacent network access, limiting exploitation to internal network segments. EPSS score of 0.12% suggests very low probability of widespread exploitation. No active exploitation or public proof-of-concept is currently known. While the vulnerability could critically compromise healthcare data within a localized network, the attack vector constraint and low EPSS reduce overall immediate risk for internet-facing environments, but it remains a high priority for healthcare providers using epa4all."

Affected products: "epa4all versions prior to 2026-05-20 (identified by CPE cpe:2.3:a:med-united:epa4all) are affected. The vendor has released a patch in version 2026-05-20, detailed in the release notes at https://github.com/med-united/epa4all/releases/tag/2026