CVE-2026-102095

Executive Summary

CVE-2026-102095 is a server-side request forgery (SSRF) vulnerability in Kiteworks Email Protection Gateway versions prior to 9.5.0. The gateway performs server-side fetches of URLs embedded in processed email message content without adequately restricting the fetch destination, allowing a remote, unauthenticated sender to force requests to internal services and cloud instance metadata endpoints. Successful exploitation can disclose sensitive internal data and, depending on the internal service reached, alter its state, yielding a CVSS score of 9.1 (Critical).

Technical Analysis

The vulnerability class is CWE-918 (Server-Side Request Forgery). The affected component is the Kiteworks Email Protection Gateway, a mail-security appliance that inspects and processes inbound email messages. According to the NVD description, the gateway "performed server-side fetches of URLs contained in the message content it processed, without adequately restricting the fetch destination."

The core defect is the absence of destination validation on attacker-influenced URLs. When a message contains a URL, the gateway dereferences that URL from its own network position. Because the gateway typically sits inside a trusted network segment with reachability to internal-only systems, the server-side fetch bypasses perimeter controls that would block an external attacker. The attacker does not need authenticated access, mailbox credentials, or user interaction — the trigger is simply delivery of a crafted message to the gateway's processing pipeline. This is consistent with the CVSS vector: AV:N (network), AC:L (low complexity), PR:N (no privileges), UI:N (no user interaction).

The description explicitly names two classes of fetch targets: (1) internal services and (2) cloud instance metadata endpoints. The request/response pattern is also significant — the gateway "return[s] the responses," meaning the fetched content is surfaced back to the party that supplied the URL. That response-return behavior converts what would otherwise be a blind SSRF into a data-exfiltration primitive. The integrity impact (I:H) derives from the description's note that exploitation may affect the state of the internal service reached, implying the forged requests are not strictly read-only.

No public proof-of-concept, exploit code, or Nuclei template is available for this CVE. The implementation-level details of the URL parser, fetch handler, and any denylist logic are not disclosed in the advisory; those specifics are therefore not characterized here. All statements above are limited to what the NVD description states directly.

Exploitation Scenarios

No public PoC is available; the attack chain below is inferred from the SSRF vulnerability class and the NVD description. No specific internal function names, endpoints, or parameters are asserted, because none are disclosed in the source data.

1. Reconnaissance / target identification. The attacker identifies an internet-facing Kiteworks Email Protection Gateway that accepts inbound mail (or otherwise ingests message content) and is running a version below 9.5.0. Version fingerprinting may rely on banner or response characteristics. The attacker does not need a valid account on the gateway.

2. Triggering the flaw. The attacker sends a message to an address processed by the gateway that contains a URL pointing at an attacker-chosen destination instead of a legitimate public resource. Because the gateway fetches URLs found in message content without adequately restricting the destination, the crafted URL causes the gateway itself to issue the request from its internal network position. No privileges or user interaction are required per the CVSS vector.

3. Payload / execution at the vulnerability-class level. The attacker selects internal targets: for example, an internal service address or a cloud instance metadata endpoint. The gateway performs the server-side fetch and — per the description — returns the fetched response to the attacker, enabling exfiltration of the retrieved content. Where the internal service supports state-changing requests, the forged request may also modify state (reflected in the I:H subscore).

4. Post-exploitation impact. Retrieved internal responses may expose sensitive data (C:H). Depending on which internal service is reached and whether it accepts state changes, the attacker may pivot further, manipulate internal service state, or harvest credentials material available from metadata endpoints. The overall blast radius is bounded by the gateway's internal network reachability and the set of services it can address.

Impact Assessment

CVSS 3.1 base score is 9.1 (Critical) with vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N. The subscores indicate: exploitation is network-reachable (AV:N), requires no special conditions (AC:L), needs no authentication (PR:N) and no user interaction (UI:N). Confidentiality and Integrity impacts are both High; Availability impact is None. Because Scope is Unchanged (S:U), impact is assessed against the vulnerable gateway's own security authority, though the internal services reached are the actual target of the forged requests.

Business impact centers on the disclosure of internal data reachable from the gateway: internal service responses and cloud instance metadata, which frequently contains temporary credentials and configuration data usable for lateral movement. Integrity impact stems from the possibility of altering the state of reached internal services. Affected systems are deployments of Kiteworks Email Protection Gateway below version 9.5.0 that process externally supplied message content. Blast radius extends to any internal or cloud-hosted service the gateway can address, making network segmentation of the gateway a decisive factor in real-world severity. No EPSS score is currently available, and the CVE is not listed in CISA KEV.

Detection Guidance

Sigma Rule

title: Kiteworks Email Protection Gateway SSRF URL Fetch Attempt
id: 7c1f2a94-3b6e-4d2a-9e51-8a0d4c6b1f23
status: experimental
description: Detects outbound HTTP fetches from the Email Protection Gateway to loopback, private, or cloud metadata destinations indicative of CVE-2026-102095 SSRF.
references:
  - https://nvd.nist.gov/vuln/detail/CVE-2026-102095
author: valtersit.com
date: 2026-01-01
logsource:
  category: proxy
detection:
  selection_process:
    process.name|contains:
      - 'kiteworks'
      - 'epg'
  selection_dest:
    destination.ip|cidr:
      - '127.0.0.0/8'
      - '10.0.0.0/8'
      - '172.16.0.0/12'
      - '192.168.0.0/16'
      - '169.254.0.0/16'
  condition: selection_process and selection_dest
falsepositives:
  - Legitimate internal health checks performed by the gateway
level: high
tags:
  - attack.t1190
  - cve.2026.102095

YARA Rule

rule CVE_2026_102095_SSRF_Metadata_Target
{
    meta:
        description = "Detects message content or logs referencing cloud instance metadata targets associated with Kiteworks EPG SSRF"
        author = "valtersit.com"
        reference = "https://nvd.nist.gov/vuln/detail/CVE-2026-102095"
        date = "2026-01-01"
        severity = "high"

    strings:
        $meta1 = "169.254.169.254" ascii wide nocase
        $meta2 = "metadata.google.internal" ascii wide nocase
        $meta3 = "metadata/computeMetadata" ascii wide nocase
        $meta4 = "latest/meta-data" ascii wide nocase
        $cve   = "CVE-2026-102095" ascii wide

    condition:
        ($cve) or (2 of ($meta*))
}

Suricata IDS Rule

alert http any any -> any any (msg:"CVE-2026-102095 Kiteworks EPG SSRF to Cloud Metadata Endpoint"; flow:established,to_server; http.host; content:"169.254.169.254"; nocase; http.uri; content:"meta-data"; nocase; classtype:attempted-recon; reference:cve,2026-102095; sid:2026102095; rev:1;)

alert http any any -> any any (msg:"CVE-2026-102095 Kiteworks EPG SSRF to Internal Address"; flow:established,to_server; http.host; pcre:"/^(127\.|10\.|172\.(1[6-9]|2[0-9]|3[01])\.|192\.168\.|169\.254\.)/"; classtype:attempted-recon; reference:cve,2026-102095; sid:2026102096; rev:1;)

Elastic Detection

sequence by host.id with maxspan=30s
  [network where process.name : ("kiteworks*", "epg*") and destination.ip : ("169.254.169.254", "127.0.0.1")]
  [http where process.name : ("kiteworks*", "epg*") and url.domain : ("metadata.google.internal", "metadata.azure.com")]

Splunk SPL Query

index=proxy OR index=firewall
| search (process_name="kiteworks*" OR process_name="epg*" OR app="kiteworks")
| eval dest_class=case(
    cidrmatch("169.254.0.0/16", dest_ip), "cloud-metadata",
    cidrmatch("127.0.0.0/8", dest_ip), "loopback",
    cidrmatch("10.0.0.0/8", dest_ip) OR cidrmatch("172.16.0.0/12", dest_ip) OR cidrmatch("192.168.0.0/16", dest_ip), "internal-rfc1918",
    true(), "other")
| where dest_class!="other"
| stats count min(_time) as first_seen max(_time) as last_seen values(dest_ip) as destinations by src_ip, dest_class
| sort - count

Wazuh OSSEC Rule

<group name="kiteworks,ssrf,">
  <rule id="100095" level="12">
    <decoded_as>json</decoded_as>
    <field name="process.name">kiteworks|epg</field>
    <field name="destination.ip">127\.0\.0\.1|169\.254\.169\.254|^10\.|^192\.168\.|^172\.(1[6-9]|2[0-9]|3[01])\.</field>
    <description>CVE-2026-102095: Kiteworks EPG server-side fetch to internal or metadata address</description>
    <mitre>
      <id>T1190</id>
    </mitre>
    <group>cve_2026_102095,</group>
  </rule>
</group>

Mitigation & Patch Guidance

Patch confidence: 0/4 (UNPATCHED — no patch evidence from any tracked source). No vendor fix or workaround is currently recorded in the available sources. The advisory states the issue is present in versions before 9.5.0, which defines the minimum fixed version but should be independently verified with the vendor before relying on it.

1. Immediate mitigations. Restrict outbound network egress from the Email Protection Gateway to only the destinations it legitimately requires. Block or tightly egress-filter access to RFC1918 ranges and the cloud instance metadata endpoint (169.254.169.254, plus provider-specific metadata hostnames) directly at the host firewall or egress gateway. Apply network segmentation so the appliance cannot freely reach internal services. Where possible, disable or limit the gateway's automatic dereferencing of URLs embedded in message content until a fix is confirmed. Enforce IMDSv2-style protections on cloud instances to reduce metadata exposure.

2. Patch/upgrade steps. Upgrade Kiteworks Email Protection Gateway to version 9.5.0 or later. Prior to upgrading in production, validate the version against the vendor's release notes, since no independent patch evidence was available at the time of writing. The appliance should be treated as unpatched until the upgrade is confirmed on all instances.

3. Long-term hardening. Maintain strict egress allowlisting for all mail-security appliances, operate them with least-privilege network reachability, and review internal services for state-changing HTTP methods callable from the gateway's segment. Monitor for anomalous fetches to internal and metadata addresses as a standing detection. Track the CVE for future vendor advisories and for any EPSS or KEV updates.

References

• NVD — CVE-2026-102095: https://nvd.nist.gov/vuln/detail/CVE-2026-102095
• CWE-918 Server-Side Request Forgery: https://cwe.mitre.org/data/definitions/918.html
• MITRE ATT&CK T1190 — Exploit Public-Facing Application: https://attack.mitre.org/techniques/T1190/
• Kiteworks product/security documentation (vendor site) for version 9.5.0 release confirmation: https://www.kiteworks.com/
• OWASP SSRF Prevention Cheat Sheet: https://cheatsheetseries.owasp.org/cheatsheets/Server_Side_Request_Forgery_Prevention_Cheat_Sheet.html

No exploit-in-the-wild evidence, threat actor campaign attribution, or public PoC is available for CVE-2026-102095 in the provided data. The CVE is not listed in CISA KEV, and no EPSS score has been published.