CWE-441
Unintended Proxy or Intermediary ('Confused Deputy')
ClassDraft
Description
The product receives a request, message, or directive from an upstream component, but the product does not sufficiently preserve the original source of the request before forwarding the request to an external actor that is outside of the product's control sphere. This causes the product to appear to be the source of the request, leading it to act as a proxy or other intermediary between the upstream component and the external actor.
Hierarchy (View 1000)
Related attack patterns (CAPEC)
CAPEC-219 · CAPEC-465
CVEs mapped to this weakness (18)
| CVE | Sev | Risk | CVSS | EPSS | KEV | Published | Description |
|---|---|---|---|---|---|---|---|
| CVE-2026-39906 | Cri | 0.65 | 10.0 | 0.00 | Apr 14, 2026 | Unisys WebPerfect Image Suite versions 3.0.3960.22810 and 3.0.3960.22604 expose a deprecated .NET Remoting TCP channel that allows remote unauthenticated attackers to leak NTLMv2 machine-account hashes by supplying a Windows UNC path as a target file argument through object-unmarshalling techniques. Attackers can capture the leaked NTLMv2 hash and relay it to other hosts to achieve privilege escalation or lateral movement depending on network configuration and patch level. | |
| CVE-2026-23751 | Cri | 0.64 | 9.8 | 0.00 | Apr 23, 2026 | Kofax Capture, now referred to as Tungsten Capture, version 6.0.0.0 (other versions may be affected) exposes a deprecated .NET Remoting HTTP channel on port 2424 via the Ascent Capture Service that is accessible without authentication and uses a default, publicly known endpoint identifier. An unauthenticated remote attacker can exploit .NET Remoting object unmarshalling techniques to instantiate a remote System.Net.WebClient object and read arbitrary files from the server filesystem, write attacker-controlled files to the server, or coerce NTLMv2 authentication to an attacker-controlled host, enabling sensitive credential disclosure, denial of service, remote code execution, or lateral movement depending on service account privileges and network environment. | |
| CVE-2025-68667 | Cri | 0.64 | — | 0.00 | Dec 23, 2025 | Conduit is a chat server powered by Matrix. A vulnerability that affects a number of Conduit-derived homeservers allows a remote, unauthenticated attacker to force the target server to cryptographically sign arbitrary membership events. Affected products include Conduit prior to version 0.10.10, continuwuity prior to version 0.5.0, Grapevine prior to commit `9a50c244`, and tuwunel prior to version 1.4.8. The flaw exists because the server fails to validate the origin of a signing request, provided the event's state_key is a valid user ID belonging to the target server. Attackers can forge "leave" events for any user on the target server. This forcibly removes users (including admins and bots) from rooms. This allows denial of service and/or the removal of technical protections for a room (including policy servers, if all users on the policy server are removed). Attackers can forge "invite" events from a victim user to themselves, provided they have an account on a server where there is an account that has the power level to send invites. This allows the attacker to join private or invite-only rooms accessible by the victim, exposing confidential conversation history and room state. Attackers can forge "ban" events from a victim user to any user below the victim user's power level, provided the victim has the power level to issue bans AND the target of the ban resides on the same server as the victim. This allows the attacker to ban anyone in a room who is on the same server as the vulnerable one, however cannot exploit this to ban users on other servers or the victim themself. Conduit fixes the issue in version 0.10.10. continuwuity fixes the issue in commits `7fa4fa98` and `b2bead67`, released in 0.5.0. tuwunel fixes the issue in commit `dc9314de1f8a6e040c5aa331fe52efbe62e6a2c3`, released in 1.4.8. Grapevine fixes the issue in commit `9a50c2448abba6e2b7d79c64243bb438b351616c`. As a workaround, block access to the `PUT /_matrix/federation/v2/invite/{roomId}/{eventId}` endpoint using your reverse proxy. | |
| CVE-2025-64125 | Cri | 0.61 | — | 0.00 | Jan 3, 2026 | A vulnerability in Nuvation Energy nCloud VPN Service allowed Network Boundary Bridging.This issue affected the nCloud VPN Service and was fixed on 2025-12-1 (December, 2025). End users do not have to take any action to mitigate the issue. | |
| CVE-2026-24471 | Cri | 0.60 | — | 0.00 | Feb 2, 2026 | continuwuity is a Matrix homeserver written in Rust. This vulnerability allows an attacker with a malicious remote server to cause the local server to sign an arbitrary event upon user interaction. Upon a user account leaving a room (rejecting an invite), joining a room or knocking on a room, the victim server may ask a remote server for assistance. If the victim asks the attacker server for assistance the attacker is able to provide an arbitrary event, which the victim will sign and return to the attacker. For the /leave endpoint, this works for any event with a supported room version, where the origin and origin_server_ts is set by the victim. For the /join endpoint, an additionally victim-set content field in the format of a join membership is needed. For the /knock endpoint, an additional victim-set content field in the format of a knock membership and a room version not between 1 and 6 is needed. This was exploited as a part of a larger chain against the continuwuity.org homeserver. This vulnerability affects all Conduit-derived servers. This vulnerability is fixed in Continuwuity 0.5.1, Conduit 0.10.11, Grapevine 0aae932b, and Tuwunel 1.4.9. | |
| CVE-2026-7381 | Cri | 0.59 | 9.1 | 0.00 | Apr 29, 2026 | Plack::Middleware::XSendfile versions through 1.0053 for Perl can allow client-controlled path rewriting. Plack::Middleware::XSendfile allows the variation setting (sendfile type) to be set by the client via the X-Sendfile-Type header, if it is not considered in the middleware constructor or the Plack environment. A malicious client can set the X-Sendfile-Type header to "X-Accel-Redirect" to services running behind nginx reverse proxies, and then set the X-Accel-Mapping to map the path to an arbitrary file on the server. Since 1.0053, Plack::Middleware::XSendfile is deprecated and will be removed from future releases of Plack. This is similar to CVE-2025-61780 for Rack::Sendfile, although Plack::Middleware::XSendfile has some mitigations that disallow regular expressions to be used in the mapping, and only apply the mapping for the "X-Accel-Redirect" type. | |
| CVE-2025-62718 | Cri | 0.57 | 9.9 | 0.00 | Apr 9, 2026 | Axios is a promise based HTTP client for the browser and Node.js. Prior to 1.15.0 and 0.31.0, Axios does not correctly handle hostname normalization when checking NO_PROXY rules. Requests to loopback addresses like localhost. (with a trailing dot) or [::1] (IPv6 literal) skip NO_PROXY matching and go through the configured proxy. This goes against what developers expect and lets attackers force requests through a proxy, even if NO_PROXY is set up to protect loopback or internal services. This issue leads to the possibility of proxy bypass and SSRF vulnerabilities allowing attackers to reach sensitive loopback or internal services despite the configured protections. This vulnerability is fixed in 1.15.0 and 0.31.0. | |
| CVE-2025-11393 | Hig | 0.57 | 8.7 | 0.00 | Dec 15, 2025 | A flaw was found in runtimes-inventory-rhel8-operator. An internal proxy component is incorrectly configured. Because of this flaw, the proxy attaches the cluster's main administrative credentials to any command it receives, instead of only the specific reports it is supposed to handle. This allows a standard user within the cluster to send unauthorized commands to the management platform, effectively acting with the full permissions of the cluster administrator. This could lead to unauthorized changes to the cluster's configuration or status on the Red Hat platform. | |
| CVE-2026-42043 | Hig | 0.47 | 7.2 | 0.00 | Apr 24, 2026 | Axios is a promise based HTTP client for the browser and Node.js. Prior to 1.15.1 and 0.31.1, an attacker who can influence the target URL of an Axios request can use any address in the 127.0.0.0/8 range (other than 127.0.0.1) to completely bypass the NO_PROXY protection. This vulnerability is due to an incomplete for CVE-2025-62718, This vulnerability is fixed in 1.15.1 and 0.31.1. | |
| CVE-2023-31313 | Hig | 0.47 | 7.2 | 0.00 | Feb 12, 2026 | An unintended proxy or intermediary in the AMD power management firmware (PMFW) could allow a privileged attacker to send malformed messages to the system management unit (SMU) potentially resulting in arbitrary code execution. | |
| CVE-2025-47269 | Hig | 0.47 | 8.3 | 0.00 | May 9, 2025 | code-server runs VS Code on any machine anywhere through browser access. Prior to version 4.99.4, a maliciously crafted URL using the proxy subpath can result in the attacker gaining access to the session token. Failure to properly validate the port for a proxy request can result in proxying to an arbitrary domain. The malicious URL `https://<code-server>/proxy/test@evil.com/path` would be proxied to `test@evil.com/path` where the attacker could exfiltrate a user's session token. Any user who runs code-server with the built-in proxy enabled and clicks on maliciously crafted links that go to their code-server instances with reference to /proxy. Normally this is used to proxy local ports, however the URL can reference the attacker's domain instead, and the connection is then proxied to that domain, which will include sending cookies. With access to the session cookie, the attacker can then log into code-server and have full access to the machine hosting code-server as the user running code-server. This issue has been patched in version 4.99.4. | |
| CVE-2025-23217 | Hig | 0.47 | — | 0.04 | Feb 6, 2025 | mitmproxy is a interactive TLS-capable intercepting HTTP proxy for penetration testers and software developers and mitmweb is a web-based interface for mitmproxy. In mitmweb 11.1.1 and below, a malicious client can use mitmweb's proxy server (bound to `*:8080` by default) to access mitmweb's internal API (bound to `127.0.0.1:8081` by default). In other words, while the cannot access the API directly, they can access the API through the proxy. An attacker may be able to escalate this SSRF-style access to remote code execution. The mitmproxy and mitmdump tools are unaffected. Only mitmweb is affected. This vulnerability has been fixed in mitmproxy 11.1.2 and above. Users are advised to upgrade. There are no known workarounds for this vulnerability. | |
| CVE-2026-27124 | Med | 0.40 | 6.1 | 0.00 | Apr 3, 2026 | FastMCP is the standard framework for building MCP applications. Prior to version 3.2.0, while testing the GitHubProvider OAuth integration, which allows authentication to a FastMCP MCP server via a FastMCP OAuthProxy using GitHub OAuth, it was discovered that the FastMCP OAuthProxy does not properly validate the user's consent upon receiving the authorization code from GitHub. In combination with GitHub’s behavior of skipping the consent page for previously authorized clients, this introduces a Confused Deputy vulnerability. This issue has been patched in version 3.2.0. | |
| CVE-2025-25061 | Med | 0.38 | 5.8 | 0.01 | Apr 4, 2025 | Unintended proxy or intermediary ('Confused Deputy') issue exists in HMI ViewJet C-more series and HMI GC-A2 series, which may allow a remote unauthenticated attacker to use the product as an intermediary for FTP bounce attack. | |
| CVE-2026-39961 | Med | 0.37 | 6.8 | 0.00 | Apr 9, 2026 | Aiven Operator allows you to provision and manage Aiven Services from your Kubernetes cluster. From 0.31.0 to before 0.37.0, a developer with create permission on ClickhouseUser CRDs in their own namespace can exfiltrate secrets from any other namespace — production database credentials, API keys, service tokens — with a single kubectl apply. The operator reads the victim's secret using its ClusterRole and writes the password into a new secret in the attacker's namespace. The operator acts as a confused deputy: its ServiceAccount has cluster-wide secret read/write (aiven-operator-role ClusterRole), and it trusts user-supplied namespace values in spec.connInfoSecretSource.namespace without validation. No admission webhook enforces this boundary — the ServiceUser webhook returns nil, and no ClickhouseUser webhook exists. This vulnerability is fixed in 0.37.0. | |
| CVE-2026-41365 | Med | 0.28 | 5.4 | 0.00 | Apr 28, 2026 | OpenClaw before 2026.3.31 contains a sender allowlist bypass vulnerability in MS Teams thread history fetched via Graph API. Attackers can retrieve thread messages that should be filtered by sender allowlists, bypassing message filtering restrictions. | |
| CVE-2026-6993 | Med | 0.27 | 5.3 | 0.00 | Apr 25, 2026 | A security flaw has been discovered in go-kratos kratos up to 2.9.2. This impacts the function NewServer of the file transport/http/server.go of the component http.DefaultServeMux Fallback Handler. The manipulation results in unintended intermediary. The attack may be launched remotely. The exploit has been released to the public and may be used for attacks. The patch is identified as 0284a5bcf92b5a7ee015300ce3051baf7ae4718d. Applying a patch is advised to resolve this issue. | |
| CVE-2025-48710 | Med | 0.20 | 4.1 | 0.01 | Jun 4, 2025 | kro (Kube Resource Orchestrator) 0.1.0 before 0.2.1 allows users (with permission to create or modify ResourceGraphDefinition resources) to supply arbitrary container images. This can lead to a confused-deputy scenario where kro's controllers deploy and run attacker-controlled images, resulting in unauthenticated remote code execution on cluster nodes. |