In its design for automatic terminal command execution, SakaDev offers two options: Execute safe commands and execute all commands. The description for the former states that commands determined by the model to be safe will be automatically executed, whereas if the model judges a command to be potentially destructive, it still requires user approval. However, this design is highly susceptible to prompt injection attacks. An attacker can employ a generic template to wrap any malicious command and mislead the model into misclassifying it as a 'safe' command, thereby bypassing the user approval requirement and resulting in arbitrary command execution.

Nette versions before 2.0.19, 2.1.13, 2.2.10, 2.3.14, 2.4.16, 3.0.6 are vulnerable to an code injection attack by passing specially formed parameters to URL that may possibly leading to RCE. Nette is a PHP/Composer MVC Framework.

In Draytek Vigor3900 1.5.1.3, attackers can inject malicious commands into mainfunction.cgi and execute arbitrary commands by calling the doGRETunnel function.

An issue in wgcloud v.2.3.7 and before allows a remote attacker to execute arbitrary code via the test connection function

Groovy Code Injection & SpEL Injection which lead to Remote Code Execution. This issue affected Apache ShenYu 2.4.0 and 2.4.1.

MetInfo CMS versions 7.9, 8.0, and 8.1 contain an unauthenticated PHP code injection vulnerability that allows remote attackers to execute arbitrary code by sending crafted requests with malicious PHP code. Attackers can exploit insufficient input neutralization in the execution path to achieve remote code execution and gain full control over the affected server.

ESPCMS P8.21120101 was discovered to contain a remote code execution (RCE) vulnerability in the component INPUT_ISDESCRIPTION.

The TELNET service in Mobatek MobaXterm 10.4 does not require authentication, which allows remote attackers to execute arbitrary commands via TCP port 23.

A vulnerability was identified in stata-mcp prior to v1.13.0 where insufficient validation of user-supplied Stata do-file content can lead to command execution.

ESPCMS P8.21120101 was discovered to contain a remote code execution (RCE) vulnerability in the component IS_GETCACHE.

RPi-Jukebox-RFID v2.7.0 was discovered to contain a remote code execution (RCE) vulnerability via htdocs\api\playlist\appendFileToPlaylist.php

The SPIP tickets plugin versions prior to 4.3.3 contain an unauthenticated remote code execution vulnerability in the forum preview handling for public ticket pages. The plugin appends untrusted request parameters into HTML that is later rendered by a template using unfiltered environment rendering (#ENV**), which disables SPIP output filtering. As a result, an unauthenticated attacker can inject crafted content that is evaluated through SPIP's template processing chain, leading to execution of code in the context of the web server.

Remote Code Execution vulnerability in DedeCMS through 5.7.109 allows remote attackers to run arbitrary code via crafted POST request to /dede/tpl.php.

Sabberworm PHP CSS Parser before 8.3.1 calls eval on uncontrolled data, possibly leading to remote code execution if the function allSelectors() or getSelectorsBySpecificity() is called with input from an attacker.

Langflow is a tool for building and deploying AI-powered agents and workflows. Prior to version 1.8.0, the CSV Agent node in Langflow hardcodes `allow_dangerous_code=True`, which automatically exposes LangChain’s Python REPL tool (`python_repl_ast`). As a result, an attacker can execute arbitrary Python and OS commands on the server via prompt injection, leading to full Remote Code Execution (RCE). Version 1.8.0 fixes the issue.

In CMS Made Simple 2.1.6, there is Server-Side Template Injection via the cntnt01detailtemplate parameter.

For RocketMQ versions 5.1.0 and below, under certain conditions, there is a risk of remote command execution.  Several components of RocketMQ, including NameServer, Broker, and Controller, are leaked on the extranet and lack permission verification, an attacker can exploit this vulnerability by using the update configuration function to execute commands as the system users that RocketMQ is running as. Additionally, an attacker can achieve the same effect by forging the RocketMQ protocol content.  To prevent these attacks, users are recommended to upgrade to version 5.1.1 or above for using RocketMQ 5.x or 4.9.6 or above for using RocketMQ 4.x .

thumbler through 1.1.2 allows OS command injection via the input, output, time, or size parameter in the thumbnail() function because user input is concatenated into a shell command string passed to child_process.exec() without proper sanitization or escaping.

MajorDoMo (aka Major Domestic Module) allows unauthenticated remote code execution via the admin panel's PHP console feature. An include order bug in modules/panel.class.php causes execution to continue past a redirect() call that lacks an exit statement, allowing unauthenticated requests to reach the ajax handler in inc_panel_ajax.php. The console handler within that file passes user-supplied input from GET parameters (via register_globals) directly to eval() without any authentication check. An attacker can execute arbitrary PHP code by sending a crafted GET request to /admin.php with ajax_panel, op, and command parameters.

Hidden functionality vulnerability in LAN-W300N/RS all versions, and LAN-W300N/PR5 all versions allows an unauthenticated attacker to log in to the product's certain management console and execute arbitrary OS commands.

textract through 2.5.0 is vulnerable to OS Command Injection via the file path parameter in multiple extractors. When processing files with malicious filenames, the filePath is passed directly to child_process.exec() in lib/extractors/doc.js, rtf.js, dxf.js, images.js, and lib/util.js with inadequate sanitization

pdf-image (npm package) through version 2.0.0 allows OS command injection via the pdfFilePath parameter. The constructGetInfoCommand and constructConvertCommandForPage functions use util.format() to interpolate user-controlled file paths into shell command strings that are executed via child_process.exec()

SQLite JDBC is a library for accessing and creating SQLite database files in Java. Sqlite-jdbc addresses a remote code execution vulnerability via JDBC URL. This issue impacting versions 3.6.14.1 through 3.41.2.1 and has been fixed in version 3.41.2.2.

An issue in Twenty CRM v1.15.0 and before allows a remote attacker to execute arbitrary code via the local.driver.ts module.

vm2 is an open source vm/sandbox for Node.js. Prior to version 3.11.0, SuppressedError allows attackers to escape the sandbox and run arbitrary code. This issue has been patched in version 3.11.0.

Orval generates type-safe JS clients (TypeScript) from any valid OpenAPI v3 or Swagger v2 specification. Versions starting with 7.19.0 and prior to 7.21.0 and 8.2.0 have an incomplete fix for CVE-2026-23947. While the jsStringEscape function properly handles single quotes ('), double quotes (") and so on, it is still possible to achieve code injection using only a limited set of characters that are currently not escaped. The vulnerability lies in the fact that the application can be forced to execute arbitrary JavaScript using characters such as []()!+. By using a technique known as JSFuck, an attacker can bypass the current sanitization logic and run arbitrary code without needing any alphanumeric characters or quotes. Version 7.21.0 and 8.2.0 contain an updated fix.

Summit is a node web framework. When using the PouchDB driver in the module, Summit 0.1.0 and later allows an attacker to execute arbitrary commands via the collection name.

CodeIgniter is a PHP full-stack web framework. This vulnerability allows attackers to execute arbitrary code when you use Validation Placeholders. The vulnerability exists in the Validation library, and validation methods in the controller and in-model validation are also vulnerable because they use the Validation library internally. This issue is patched in version 4.3.5.

vm2 is an open source vm/sandbox for Node.js. Prior to version 3.11.0, VM2 suffers from a sandbox breakout vulnerability through the inspect function. This allows attackers to write code which can escape from the VM2 sandbox and execute arbitrary commands on the host system. This issue has been patched in version 3.11.0.

The Zabbix Agent 2 item key smart.disk.get does not sanitize its parameters before passing them to a shell command resulting possible vulnerability for remote code execution.

An issue was discovered in goform/formsetUsbUnload in Tenda AC15V1.0 V15.03.05.18_multi. The value of `v1` was not checked, potentially leading to a command injection vulnerability if injected into doSystemCmd.

vm2 is an open source vm/sandbox for Node.js. Prior to version 3.10.5, the fix for CVE-2023-37466 is insufficient and can be circumvented allowing attackers to write code which can escape from the VM2 sandbox and execute arbitrary commands on the host system. This issue has been patched in version 3.10.5.

Orval generates type-safe JS clients (TypeScript) from any valid OpenAPI v3 or Swagger v2 specification. Versions 7.19.0 and below and 8.0.0-rc.0 through 8.0.2 allow untrusted OpenAPI specifications to inject arbitrary TypeScript/JavaScript into generated mock files via the const keyword on schema properties. These const values are interpolated into the mock scalar generator (getMockScalar in packages/mock/src/faker/getters/scalar.ts) without proper escaping or type-safe serialization, which results in attacker-controlled code being emitted into both interface definitions and faker/MSW handlers. The vulnerability is similar in impact to the previously reported enum x-enumDescriptions (GHSA-h526-wf6g-67jv), but it affects a different code path in the faker-based mock generator rather than @orval/core. The issue has been fixed in versions 7.20.0 and 8.0.3.

Orval generates type-safe JS clients (TypeScript) from any valid OpenAPI v3 or Swagger v2 specification. Versions prior to 7.19.0 until 8.0.2 are vulnerable to arbitrary code execution in environments consuming generated clients. This issue is similar in nature to CVE-2026-22785, but affects a different code path in @orval/core that was not addressed by CVE-2026-22785's fix. The vulnerability allows untrusted OpenAPI specifications to inject arbitrary TypeScript/JavaScript code into generated clients via the x-enumDescriptions field, which is embedded without proper escaping in getEnumImplementation(). I have confirmed that the injection occurs during const enum generation and results in executable code within the generated schema files. Orval 7.19.0 and 8.0.2 contain a fix for the issue.

An issue was discovered in Tenda W20E V4.0br_V15.11.0.6. Failure to validate the value of `usbPartitionName`, which is directly used in `doSystemCmd`, may lead to critical command injection vulnerabilities.

vm2 is an open source vm/sandbox for Node.js. Prior to version 3.11.0, VM2 suffers from a sandbox breakout vulnerability. This allows attackers to write code which can escape from the VM2 sandbox and execute arbitrary commands on the host system. This issue has been patched in version 3.11.0.

Eigent is a multi-agent Workforce. A critical security vulnerability in the CI workflow (.github/workflows/ci.yml) allows arbitrary code execution from fork pull requests with repository write permissions. The vulnerable workflow uses pull_request_target trigger combined with checkout of untrusted PR code. An attacker can exploit this to steal credentials, post comments, push code, or create releases.

Improper Control of Generation of Code ('Code Injection') vulnerability in Salesforce Uni2TS on MacOS, Windows, Linux allows Leverage Executable Code in Non-Executable Files.This issue affects Uni2TS: through 1.2.0.

Cursor is a code editor built for programming with AI. Prior to 2.3, hen the Cursor Agent is running in Auto-Run Mode with Allowlist mode enabled, certain shell built-ins can still be executed without appearing in the allowlist and without requiring user approval. This allows an attacker via indirect or direct prompt injection to poison the shell environment by setting, modifying, or removing environment variables that influence trusted commands. This vulnerability is fixed in 2.3.

CrewAI does not properly check that Docker is still running during runtime, and will fall back to a sandbox setting that allows for RCE exploitation.

The CloudStack integration API service allows running its unauthenticated API server (usually on port 8096 when configured and enabled via integration.api.port global setting) for internal portal integrations and for testing purposes. By default, the integration API service port is disabled and is considered disabled when integration.api.port is set to 0 or negative. Due to an improper initialisation logic, the integration API service would listen on a random port when its port value is set to 0 (default value). An attacker that can access the CloudStack management network could scan and find the randomised integration API service port and exploit it to perform unauthorised administrative actions and perform remote code execution on CloudStack managed hosts and result in complete compromise of the confidentiality, integrity, and availability of CloudStack managed infrastructure. Users are recommended to restrict the network access on the CloudStack management server hosts to only essential ports. Users are recommended to upgrade to version 4.18.2.1, 4.19.0.2 or later, which addresses this issue.

vm2 is an open source vm/sandbox for Node.js. In vm2 prior to version 3.10.2, `Promise.prototype.then` `Promise.prototype.catch` callback sanitization can be bypassed. This allows attackers to escape the sandbox and run arbitrary code. In lib/setup-sandbox.js, the callback function of `localPromise.prototype.then` is sanitized, but `globalPromise.prototype.then` is not sanitized. The return value of async functions is `globalPromise` object. Version 3.10.2 fixes the issue.

vLLM is an inference and serving engine for large language models (LLMs). Starting in version 0.10.1 and prior to version 0.14.0, vLLM loads Hugging Face `auto_map` dynamic modules during model resolution without gating on `trust_remote_code`, allowing attacker-controlled Python code in a model repo/path to execute at server startup. An attacker who can influence the model repo/path (local directory or remote Hugging Face repo) can achieve arbitrary code execution on the vLLM host during model load. This happens before any request handling and does not require API access. Version 0.14.0 fixes the issue.

Improper Control of Generation of Code ('Code Injection') vulnerability in Johnson Controls Frick Controls Quantum HD allows Code Injection. Insufficient validation of input in certain parameters may permit unexpected actions, which could impact the security of the device before authentication occurs.This issue affects Frick Controls Quantum HD version 10.22 and prior.

Improper Control of Generation of Code ('Code Injection') vulnerability in Johnson Controls Frick Controls Quantum HD allows Code Injection. Insufficient validation of input in certain parameters may permit unexpected actions, which could impact the security of the device before authentication occurs.This issue affects Frick Controls Quantum HD version 10.22 and prior.

In Advantech WebAccss/SCADA v9.1.3 and prior, there is an arbitrary file overwrite vulnerability, which could allow an attacker to overwrite any file in the operating system (including system files), inject code into an XLS file, and modify the file extension, which could lead to arbitrary code execution.

A code injection in Ivanti Endpoint Manager Mobile allowing attackers to achieve unauthenticated remote code execution.

user_login.cgi on Draytek Vigor2620 devices before 3.9.8.4 (and on all versions of Vigor2925 devices) allows attackers to send a crafted payload to modify the content of the code segment, insert shellcode, and execute arbitrary code.

The Score extension through 0.3.0 for MediaWiki has a remote code execution vulnerability due to improper sandboxing of the GNU LilyPond executable. This allows any user with an ability to edit articles (potentially including unauthenticated anonymous users) to execute arbitrary Scheme or shell code by using crafted {{Image data to generate musical scores containing malicious code.

In Xerox FreeFlow Core version 8.0.4, an attacker can exploit a Path Traversal vulnerability to access unauthorized files on the server. This can lead to Remote Code Execution (RCE), allowing the attacker to run arbitrary commands on the system.