The internet is broken right now.
65 critical CVEs without a fix in the last 48 hours. Developers are crying quietly into their IDEs. 146 more already have a fix that someone still has to apply.
-
CVE-2025-43874
: Improper Neutralization of Special Elements used in an OS Command vulnerability in Johnson Controls iSTAR Ultra, Johnson Controls iSTAR Ultra SE, Johnson Controls iSTAR Ultra G2, Johnson Controls iSTAR Ultra G2 SE, and Johnson Controls iSTAR Edge G2 allows : OS Command Injection. This issue affects iSTAR Ultra: before 6.9.2.CU02; iSTAR Ultra SE: before 6.9.2.CU02; iSTAR Ultra G2: before 6.9.2.CU02; iSTAR Ultra G2 SE: before 6.9.2.CU02; iSTAR Edge G2: before 6.9.2.CU02.
-
CVE-2026-82912
Improper Neutralization of Special Elements Used in an Expression Language Statement ('Expression Language Injection') in EmailServices in Apache OFBiz 24.09 before 24.09.08 on all platforms allows a remote authenticated attacker holding only OFBTOOLS_VIEW and ACCOUNTING_VIEW, the two view-level permissions that open the Accounting application, to execute arbitrary operating system commands as the account running OFBiz, by submitting a ${groovy:...} expression as the bodyText or subject value to the shipped Send Per Email invoice form, because sendMailFromScreen passes those caller-supplied values through FlexibleStringExpander.expandString, which compiles and evaluates embedded Groovy on a class loader that carries none of the SecureASTCustomizer restrictions GroovyUtil.eval applies elsewhere, behind a service and request-map that declare no permission check beyond base authentication. Users are recommended to upgrade to version 24.09.08, which fixes this issue.
-
CVE-2026-126080
Active Support is a collection of utility classes and standard library extensions used by Rails and other Ruby projects. Prior to 8.1.4.1 and 8.0.5.2, when constant lookup raises LoadError, ActiveSupport::Inflector#safe_constantize (and String#safe_constantize) builds a regular expression from the requested constant name without escaping it, so a caller that passes untrusted input lets the attacker choose the source of a pattern that the application compiles. Crafted patterns can trigger defects in Ruby's regular expression compiler, possibly leading to a crash of the server process, and repeated requests could keep the application unavailable. Action Cable passes the channel value of each subscribe command to safe_constantize before any channel lookup, so an unauthenticated client can reach this code when the application mounts Action Cable and accepts connections from untrusted clients without an added authentication check. This issue is fixed in versions 8.1.4.1 and 8.0.5.2.
-
CVE-2026-83953
Improper neutralization of special elements used in a command ('command injection') in Github Copilot allows an authorized attacker to disclose information locally.
-
CVE-2026-125757
mailcow: dockerized before 2026-11 contains a server-side template injection vulnerability that allows domain administrators to execute commands in the shared rspamd container by storing a malicious domain-wide footer. Attackers can embed a Jinja macro with an os.execute default value, rendered by the rspamd MOO_FOOTER prefilter, to read or inject mail for every tenant and disable filtering.
-
CVE-2026-105669
Argument injection in Ivanti Sentry before version 11.0, 10.8.3, and 10.7.4 allows a remote authenticated attacker with admin privileges to execute arbitrary on-disk binaries.
-
CVE-2026-123387
Improper neutralization of special elements used in a command ('command injection') in Windows Active Directory allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-82910
Improper Neutralization of Special Elements Used in a Template Engine ('Server-Side Template Injection') in MacroScreenRenderer in Apache OFBiz 24.09 before 24.09.08 on all platforms allows a remote authenticated attacker holding only the two view permissions of a control-servlet webapp (e.g. CATALOG_VIEW and OFBTOOLS_VIEW on /catalog, held by OFBiz's own shipped order-entry clerk role) to execute arbitrary operating system commands as the OFBiz service account with a single GET request, by placing an apostrophe followed by FreeMarker expression source in a screen link parameter value, because renderLink concatenates link parameter values verbatim between apostrophes into an unquoted, unescaped template-source string that is compiled and evaluated at render time, allowing the injected expression to reach the allow-listed FlexibleStringExpander.expandString call and, through a runtime-assembled Groovy interpolation, unsandboxed code execution. Users are recommended to upgrade to version 24.09.08, which fixes this issue.
-
CVE-2026-97693
Yuxi is a large-model-based intelligent knowledge base and knowledge graph agent development platform. From 0.6.0.beta1 until 0.7.1, there is insufficient isolation between execution-capable Docker sandboxes and trusted application services. A low-privileged authenticated user who can execute commands in their own sandbox can reach other tenants' sandbox APIs, access their files and secrets, execute commands in active sandboxes, and enumerate, create, or delete sandboxes through the unauthenticated provisioner without knowing administrator, database, or MinIO credentials. Sandboxes can also reach shared internal services; if shipped default service credentials remain unchanged, an attacker can use them, and if the exposed PostgreSQL account is a superuser, obtain command execution in the PostgreSQL container. These paths can disclose or modify cross-tenant data and sandbox state or cause denial of service through sandbox deletion without a container escape or kernel vulnerability. This vulnerability is fixed in 0.7.1. Existing deployments must separately rotate persisted PostgreSQL, Neo4j, or MinIO credentials that were previously set to documented or shipped defaults because upgrading the Compose files does not automatically rotate them.
-
CVE-2026-110193
IBM Langflow OSS 1.0.0 through 1.12.8 could allow a remote attacker with a low-privileged authenticated account to cause the Agentic Assistant to generate or submit component code using these unblocked patterns, resulting in arbitrary operating system command execution on the server.
-
CVE-2026-126224
A improper neutralization of argument delimiters in a command ('argument injection') vulnerability in Fortinet FortiAnalyzer 8.0.0, FortiAnalyzer 7.6.0 through 7.6.7, FortiAnalyzer 7.4.0 through 7.4.11, FortiAnalyzer 7.2.0 through 7.2.12, FortiAnalyzer Cloud 8.0.0, FortiAnalyzer Cloud 7.6.0 through 7.6.7, FortiAnalyzer Cloud 7.4.1 through 7.4.11, FortiAnalyzer Cloud 7.2.1 through 7.2.12, FortiManager 8.0.0, FortiManager 7.6.0 through 7.6.7, FortiManager 7.4.0 through 7.4.11, FortiManager 7.2.0 through 7.2.12, FortiManager Cloud 8.0.0, FortiManager Cloud 7.6.0 through 7.6.7, FortiManager Cloud 7.4.1 through 7.4.11, FortiManager Cloud 7.2.1 through 7.2.12 may allow an authenticated privileged admin user to execute unauthorized code or commands.
-
CVE-2026-119906
Improper neutralization of special elements used in an os command ('os command injection') in Windows Sandbox allows an authorized attacker to elevate privileges locally.
-
CVE-2026-119713
Paperclip 2026.403.0 before 2026.416.0 contains a remote code execution vulnerability that allows agent API key holders to set cleanupCommand and teardownCommand via PATCH /api/execution-workspaces/:id. When the workspace is archived, cleanupExecutionWorkspaceArtifacts runs these commands through a shell as the Paperclip server process, executing attacker-chosen OS commands on the server host.
-
CVE-2026-114720
Roundcube Webmail before 1.6.20 and 1.7.x before 1.7.5 is affected by an IMAP command injection vulnerability. A remote authenticated attacker can exploit this via CRLF sequences injected into the _flag parameter of the mail/mark action to execute arbitrary IMAP commands. Upstream fixed this in the commit “Fix IMAP command injection via CRLF in `_flag` parameter of mail/mark action”, which rejects flags containing control, whitespace and IMAP-special characters.
-
CVE-2026-82960
Improper Control of Generation of Code ('Code Injection') in ModelFormField in the Apache OFBiz cms plugin before 24.09.08 on all platforms allows a remote authenticated attacker holding only CMS content-editor access to execute arbitrary operating system commands as the OFBiz process user via a single HTTP POST request, by submitting a crafted caContentIdTo parameter to EditAddContent, because the CMS screen copies that parameter into a top-level context variable that is interpolated into a form field's use-when attribute and the resulting string is passed unvalidated to GroovyUtil.eval(), whose compile-time SecureASTCustomizer sandbox does not restrict runtime reflection, allowing a payload that closes the expanded string literal and invokes java.lang.Runtime via reflection. Users are recommended to upgrade to version 24.09.08, which fixes this issue.
-
CVE-2026-77880
Control Web Panel (CWP) before webpanel_version 1.16 contains an OS command injection vulnerability in the loopback API password change route /newpass/, which runs as root. An attacker can inject arbitrary OS commands by manipulating the POST 'user' parameter that is passed unescaped into a shell command, achieving remote code execution with root privileges.
-
CVE-2026-120275
A security flaw has been discovered in Shanghai Xi Yu Ji Zhi Technology Minimax Code 0.5.8. This affects an unknown part of the component Find Command Handler. The manipulation of the argument -exec results in os command injection. The attack may be performed from remote. The vendor was contacted early about this disclosure but did not respond in any way.
-
CVE-2026-118114
Improper neutralization of special elements used in an sql command ('sql injection') in SQL Server allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-118115
Improper neutralization of special elements used in an sql command ('sql injection') in SQL Server allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-78845
Aidot products Linkind Matter Smart Plug 1.01.32, OREiN Matter Smart Light Bulb 3.15.21, and Linkind Wi-Fi Smart Light Bulb 3.30.17 are vulnerable to Remote Code Execution (RCE) via UPnP root description parsing.
-
CVE-2026-79882
There is a command injection vulnerability in the PAPI protocol of AOS-10 that could lead to unauthenticated remote code execution by an adjacent attacker sending specially crafted packets. Successful exploitation of this vulnerability could allow an attacker to execute arbitrary code as a privileged user on the underlying operating system.
-
CVE-2026-120338
Prefab UI's Generative UI sandbox runs caller-supplied Python inside Pyodide, Python compiled to WebAssembly, in a Deno subprocess. The feature is documented as isolation for untrusted code, including LLM-written UI. From prefab-ui 0.14.0 through 0.20.2 the launcher starts Deno with unrestricted --allow-read, --allow-write, and --allow-net, and the runner leaves the Deno API on the JavaScript global object. Pyodide's js module proxies that global, so the Python can call Deno.readTextFileSync and Deno.writeTextFileSync. A caller who can supply the code can read and write files as the server OS user, and can persist code execution by writing a .pth file into the server virtualenv's site-packages. The usual remote path is FastMCP's GenerativeUI provider, which registers the MCP tool generate_prefab_ui and forwards that tool's code argument into this sandbox. On a FastMCP server with no authentication configured, any client who can reach the server can call the tool. With authentication, any client allowed to call tools can. Generative UI is opt-in, and the host must have Deno installed. Confirmed on prefab-ui 0.20.2. Fixed in 0.21.0: after the runtime and Prefab sources load, the runner revokes Deno read, write, and network permission before it accepts code, the process is started with --no-prompt so a permission request cannot grant them again, and bootstrap network access is limited to cdn.jsdelivr.net. The fix is commit 030f074 (pull request 538).
-
CVE-2026-54141
OpenHuman is an open-source agent harness available as a desktop application and Rust library. Before 0.57.13, a remote direct message received through a configured Yuanbao channel is converted into a generic channel AgentTurnRequest that uses the default tool policy without establishing an interactive approval context. Under the default Supervised autonomy policy with Node enabled, node_exec is classified as a write operation that should prompt for approval, but ApprovalGate treats missing APPROVAL_CHAT_CONTEXT as a non-interactive turn and allows execution. Because Yuanbao direct messages are open by default, an external sender who can message the connected bot can influence the agent to execute JavaScript and operating-system commands as the OpenHuman user without an approval prompt. The issue requires a configured Yuanbao channel, the default approval-requiring policy state, Node availability, and successful model selection of node_exec. This issue is fixed in version 0.57.13.
-
CVE-2026-109457
Improper neutralization of special elements used in an os command ('os command injection') in Azure CLI allows an authorized attacker to execute code over a network.
-
CVE-2026-121350
Adobe Connect is affected by an Improper Neutralization of Special Elements used in an SQL Command ('SQL Injection') vulnerability that could lead to disclosure of sensitive memory. A low-privileged attacker could leverage this vulnerability to disclose sensitive information. Exploitation of this issue does not require user interaction. Scope is changed.
-
CVE-2026-130848
Dolibarr ERP CRM through 24.0.3 contains a code injection vulnerability in Website::importWebSite() that allows website editors without PHP permission to execute code by importing templates. Attackers can import a template zip whose javascript.js.php contains PHP code, which the public javascript.js.php endpoint includes, executing code as the web server user.
-
CVE-2026-61991
Unauthenticated Remote Code Execution (RCE) in AjaxLoadMore Cache <= 3.1.1 versions.
-
CVE-2026-104339
IBM Langflow OSS 1.0.0 through 1.12.3 could allow a remote attacker to execute arbitrary code due to improper neutralization of special elements used in an OS command.
-
CVE-2026-115785
An SMTP command injection vulnerability in the data-source testing functionality of Zimbra Collaboration Suite (ZCS) allows an ordinary authenticated user with account options enabled to inject CR/LF into the emailAddress inserted into an SMTP envelope command by a JSON SOAP request. After a successful preliminary POP3 or IMAP login, Zimbra sends the injected commands to the selected SMTP service. If that service accepts the pipelined transaction, the attacker can submit mail with attacker-selected recipients, headers and body content without victim interaction. The vulnerable command-construction path is present in ZCS 10.1.16 and 10.1.20.
-
CVE-2026-115904
Topluyo is a social networking platform. A vulnerability exists in Topluyo Desktop for Linux prior to version 2026.9.2410 where user-controlled URLs are unsafely processed, leading to OS Command Injection via the `openExternalLinks()` routine in `utils.js`. An attacker who controls a target URL can execute arbitrary operating system commands with the privileges of the user running the application upon opening the link. Version 2026.9.2410 patches the issue.
-
CVE-2026-110697
A flaw was found in Quay where the robot account listing function constructs a SQL LIKE pattern from user-controlled namespace names without escaping wildcard metacharacters. An attacker who creates a namespace containing an underscore character can match and retrieve robot accounts and their decrypted authentication tokens from other tenants' namespaces, leading to cross-tenant credential theft and unauthorized image operations.
-
CVE-2026-83816
CodeWhale is a DeepSeek and MiMo coding agent for the terminal. Prior to 0.9.0, the ExecShellTool::approval_requirement_for path used is_parallel_readonly_command to downgrade commands to automatic approval based on a flag-stripped command allowlist. The allowlist treated fd, rg, and git grep as read-only even when fd -x or fd --exec, rg --pre, or git grep --open-files-in-pager selected an external program. registered_tool_approval_required returned without displaying a prompt for that automatic classification, and the later command-safety check also treated the base utilities as safe. Attacker-controlled content that influenced an agent tool call could therefore cause arbitrary program execution on the developer host without the approval promised by the default interactive profile, resulting in unauthorized access, modification, or disruption. This issue is fixed in version 0.9.0.
-
CVE-2026-121428
AzuraCast through 0.23.8 contains a remote code execution vulnerability that allows authenticated station managers to inject Supervisor configuration through the station timezone field. Attackers can submit a timezone containing newlines to the profile edit API, defining a new Supervisor program that executes OS commands as root when configuration regenerates.
-
CVE-2026-115188
An unrestricted file upload vulnerability exists in the campaign import feature of Mautic 7. When processing an uploaded campaign archive, the application copies the bundled asset files into a publicly accessible media directory without restricting their file names or types. An authenticated user with permission to create campaign imports can exploit this through the user interface or the API (/api/campaigns/import) to place arbitrary files, including server-side scripts, in the web root. Administrator privileges are not required. On web servers that honor per-directory configuration, such as the default Apache setup of the official Mautic Docker image, this leads to remote code execution with the privileges of the web server user, including access to the Mautic configuration, application secrets, and contact data.
-
CVE-2026-96053
Improper neutralization of special elements used in a command ('command injection') in Windows Task Scheduler allows an authorized attacker to elevate privileges locally.
-
CVE-2026-113403
ArmorX Spam developed by ArmorX Global Technology has a Command Injection vulnerability. Authenticated remote attackers can inject crafted commands into a specific parameter, thereby uploading malicious code to the server.
-
CVE-2026-102224
Improper neutralization of special elements used in an os command ('os command injection') in Azure CycleCloud allows an authorized attacker to elevate privileges over a network.
-
CVE-2026-95756
A command injection vulnerability exists in the API of HPE Networking Fabric Composer and could be exploited by an authenticated remote attacker with administrative privileges if certain preconditions outside of the attacker's control are met. Successful exploitation could allow an attacker to gain elevated privileges on the underlying operating system.
-
CVE-2026-125637
A flaw was found in the Konflux UI proxy. Hostname values defined in the KonfluxUI custom resource are inserted directly into proxy configuration files without adequate validation or sanitization. An authenticated user with permissions to modify the custom resource can inject malicious configuration directives to bypass authentication and authorization controls, potentially exposing internal backend services. In addition, providing malformed input can cause the configuration generator to fail, resulting in a denial of service (DoS) by preventing the proxy from starting.
-
CVE-2026-123784
Fluxbox through 1.3.7 contains an OS command injection vulnerability in RootTheme::reconfigTheme() that allows attackers to execute commands via crafted style background values. Attackers can embed shell metacharacters in background pixmap paths, random directories, or gradient texture text, which are concatenated unescaped into an fbsetbg command run via $SHELL -c when the style loads.
-
CVE-2026-116692
MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. From 3.12.0rc0 until 3.17.0, third-party scorer deserialization in mlflow/genai/scorers/base.py accepts request-selected descendants of THIRD_PARTY_SCORER_ALLOWED_MODULES rather than requiring an exact allowed module. A caller can use MLflow artifact endpoints to place a Python module beneath a writable allowed package and submit its module and class through /ajax-api/3.0/mlflow/scorer/invoke. Scorer.model_validate_json() imports the module and constructs the class before trace processing. Authentication-disabled deployments permit this code execution without credentials, while basic-auth deployments require a principal capable of creating an experiment and run and uploading artifacts. In either configuration, the allowed scorer package must be writable by the server process. This issue is fixed in version 3.17.0.
-
CVE-2026-79956
A vulnerability in the CLI command handling of the AOS-8 web interface could allow an authenticated remote attacker to execute arbitrary commands on the underlying operating system. Successful exploitation could allow an attacker to run unauthorized commands on the system, potentially leading to system compromise.
-
CVE-2026-79849
Vulnerabilities in the affected interface of AOS-8 could allow an unauthenticated remote attacker to conduct SQL injection attacks against an affected AOS-8 instance. Successful exploitation could allow an attacker to execute arbitrary commands on the underlying operating system.
-
CVE-2026-79848
Vulnerabilities in the affected interface of AOS-8 could allow an unauthenticated remote attacker to conduct SQL injection attacks against an affected AOS-8 instance. Successful exploitation could allow an attacker to execute arbitrary commands on the underlying operating system.
-
CVE-2026-107326
IBM Langflow OSS 1.0.0 through 1.12.4 could allow a remote authenticated attacker to execute arbitrary code due to improper neutralization of special elements used in an OS command.
-
CVE-2026-104328
IBM Langflow OSS 1.0.0 through 1.12.3 could allow a remote authenticated attacker to execute arbitrary code due to improper neutralization of special elements used in an OS command.
-
CVE-2026-101288
IBM Langflow OSS 1.0.0 through 1.12.3 could allow a remote authenticated attacker to execute arbitrary code due to improper neutralization of special elements used in code.
-
CVE-2026-117573
hagent through commit 761eec6 contains an OS command injection vulnerability that allows MCP clients to execute shell commands by injecting metacharacters into the args parameter of the hagent_arduino and hagent_esp32 tools. Attackers or prompt-injected agents can supply crafted args interpolated into arduino-cli and idf.py commands run with subprocess.run(shell=True) to execute commands as the MCP server user.
-
CVE-2026-92207
MindsDB OpenBBtable Code Injection Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of MindsDB. Authentication is required to exploit this vulnerability. The specific flaw exists within the OpenBBtable class. The issue results from the lack of proper validation of a user-supplied string before using it to execute Python code. An attacker can leverage this vulnerability to execute code in the context of the service account. Was ZDI-CAN-28388.
-
CVE-2026-126234
A improper neutralization of special elements used in a template engine vulnerability in Fortinet FortiAnalyzer 8.0.0, FortiAnalyzer Cloud 8.0.0, FortiManager 8.0.0, FortiManager Cloud 8.0.0 may allow an authenticated attacker with restricted privileges to execute unauthorized code or commands via a specially crafted request.