Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. When Auto-GPT is executed directly on the host system via the provided run.sh or run.bat files, custom Python code execution is sandboxed using a temporary dedicated docker container which should not have access to any files outside of the Auto-GPT workspace directory. Before v0.4.3, the `execute_python_code` command (introduced in v0.4.1) does not sanitize the `basename` arg before writing LLM-supplied code to a file with an LLM-supplied name. This allows for a path traversal attack that can overwrite any .py file outside the workspace directory by specifying a `basename` such as `../../../main.py`. This can further be abused to achieve arbitrary code execution on the host running Auto-GPT by e.g. overwriting autogpt/main.py which will be executed outside of the docker environment meant to sandbox custom python code execution the next time Auto-GPT is started. The issue has been patched in version 0.4.3. As a workaround, the risk introduced by this vulnerability can be remediated by running Auto-GPT in a virtual machine, or another environment in which damage to files or corruption of the program is not a critical problem.

A vulnerability in significant-gravitas/autogpt version 0.5.1 allows an attacker to bypass the shell commands denylist settings. The issue arises when the denylist is configured to block specific commands, such as 'whoami' and '/bin/whoami'. An attacker can circumvent this restriction by executing commands with a modified path, such as '/bin/./whoami', which is not recognized by the denylist.

AutoGPT, a component of significant-gravitas/autogpt, is vulnerable to an improper neutralization of special elements used in an OS command ('OS Command Injection') due to a flaw in its shell command validation function. Specifically, the vulnerability exists in versions v0.5.0 up to but not including 5.1.0. The issue arises from the application's method of validating shell commands against an allowlist or denylist, where it only checks the first word of the command. This allows an attacker to bypass the intended restrictions by crafting commands that are executed despite not being on the allowlist or by including malicious commands not present in the denylist. Successful exploitation of this vulnerability could allow an attacker to execute arbitrary shell commands.

spice-vdagent up to and including 0.17.0 does not properly escape save directory before passing to shell, allowing local attacker with access to the session the agent runs in to inject arbitrary commands to be executed.

A vulnerability has been identified in SCALANCE LPE9403 (6GK5998-3GS00-2AC2) (All versions with SINEMA Remote Connect Edge Client installed). Affected devices do not properly sanitize configuration parameters. This could allow a non-privileged local attacker to execute root commands on the device.

An improper neutralization of special elements used in an OS command vulnerability [CWE-78] in FortiADC 7.2.0, 7.1.0 through 7.1.1 may allow an authenticated attacker to execute unauthorized commands via specifically crafted arguments to existing commands.

An improper neutralization of special elements used in an OS Command ('OS Command Injection') vulnerability [CWE-78 ] in FortiManager 7.2.0 through 7.2.2, 7.0.0 through 7.0.7, 6.4.0 through 6.4.11, 6.2 all versions, 6.0 all versions, FortiAnalyzer 7.2.0 through 7.2.2, 7.0.0 through 7.0.7, 6.4.0 through 6.4.11, 6.2 all versions, 6.0 all versions and FortiADC  7.1.0, 7.0.0 through 7.0.3, 6.2 all versions, 6.1 all versions, 6.0 all versions management interface may allow an authenticated attacker with at least READ permissions on system settings to execute arbitrary commands on the underlying shell due to an unsafe usage of the wordexp function.

All versions of the package bwm-ng are vulnerable to Command Injection due to improper input sanitization in the 'check' function in the bwm-ng.js file. **Note:** To execute the code snippet and potentially exploit the vulnerability, the attacker needs to have the ability to run Node.js code within the target environment. This typically requires some level of access to the system or application hosting the Node.js environment.

All versions of the package keep-module-latest are vulnerable to Command Injection due to missing input sanitization or other checks and sandboxes being employed to the installModule function. **Note:** To execute the code snippet and potentially exploit the vulnerability, the attacker needs to have the ability to run Node.js code within the target environment. This typically requires some level of access to the system or application hosting the Node.js environment.

Multiple improper neutralization of special elements used in an os command ('OS Command Injection') vulnerabilties [CWE-78] vulnerability in Fortinet allows a local authenticated attacker to execute arbitrary shell code as `root` user via crafted CLI requests.

Dell Unity, version(s) 5.5 and prior, contain(s) an OS Command Injection Vulnerability in its svc_nfssupport utility. An authenticated attacker could potentially exploit this vulnerability, escaping the restricted shell and execute arbitrary operating system commands with root privileges.

AppUse 4.0 allows shell command injection via a proxy field.

Dell VxRail versions earlier than 7.0.450, contain(s) an OS command injection vulnerability in VxRail Manager. A local authenticated attacker could potentially exploit this vulnerability, leading to the execution of arbitrary OS commands on the application's underlying OS, with the privileges of the vulnerable application. Exploitation may lead to a system take over by an attacker.

Command injection vulnerability in Document.php in Synology Office 2.2.0-1502 and 2.2.1-1506 allows remote authenticated users to execute arbitrary commands via shell metacharacters in the crafted file name of RTF documents.

Linux Dash up to version v2 is vulnerable to multiple command injection vulnerabilities in the way module names are parsed and then executed resulting in code execution on the server, potentially as root.

Command injection in evince via filename when printing to PDF. This affects versions earlier than 3.25.91.

IBM Security Guardium Database Activity Monitor appliance could allow a local user to inject commands that would be executed as root.

iox in Cisco IOS, possibly 15.6 and earlier, and IOS XE, possibly 3.18 and earlier, allows local users to execute arbitrary IOx Linux commands on the guest OS via crafted iox command-line options, aka Bug ID CSCuz59223.

A vulnerability classified as critical was found in KylinSoft youker-assistant on KylinOS. Affected by this vulnerability is the function adjust_cpufreq_scaling_governer. The manipulation leads to os command injection. It is possible to launch the attack on the local host. The exploit has been disclosed to the public and may be used. Upgrading to version 3.1.4.13 is able to address this issue. It is recommended to upgrade the affected component. The associated identifier of this vulnerability is VDB-226099.

A vulnerability in the CLI of Cisco NX-OS Software could allow an authenticated, local attacker to execute arbitrary commands on the underlying operating system of an affected device. This vulnerability is due to insufficient validation of arguments that are passed to specific CLI commands. An attacker could exploit this vulnerability by including crafted input as the argument of an affected command. A successful exploit could allow the attacker to execute arbitrary commands on the underlying operating system with the privileges of the currently logged-in user.

Multiple vulnerabilities in the restricted shell of Cisco Evolved Programmable Network Manager (EPNM), Cisco Identity Services Engine (ISE), and Cisco Prime Infrastructure could allow an authenticated, local attacker to escape the restricted shell and gain root privileges on the underlying operating system. For more information about these vulnerabilities, see the Details section of this advisory.

Cisco Unified Computing System (UCS) Platform Emulator 2.5(2)TS4, 3.0(2c)A, and 3.0(2c)TS9 allows local users to gain privileges via crafted arguments on a ucspe-copy command line, aka Bug ID CSCux68832.

A vulnerability, which was classified as critical, was found in kylin-system-updater up to 1.4.20kord on Ubuntu Kylin. Affected is the function InstallSnap of the component Update Handler. The manipulation leads to command injection. The attack needs to be approached locally. The exploit has been disclosed to the public and may be used. The identifier of this vulnerability is VDB-222600.

Dell Unity, version(s) 5.4 and prior, contain(s) an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Command execution and Elevation of privileges.

Dell Unity, version(s) 5.4 and prior, contain(s) an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Code execution and Elevation of privileges.

Dell Unity, version(s) 5.4 and prior, contain(s) an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Command execution and Elevation of privileges.

Dell Unity, version(s) 5.5 and prior, contain(s) an OS Command Injection Vulnerability in its svc_nas utility. An authenticated attacker could potentially exploit this vulnerability, escaping the restricted shell and execute arbitrary operating system commands with root privileges.

SOUND4 IMPACT/FIRST/PULSE/Eco <=2.x contains a conditional command injection vulnerability that allows local authenticated users to create malicious files in the /tmp directory. Unauthenticated attackers can execute commands by making a single HTTP POST request to the traceroute.php script, which triggers the malicious file and then deletes it after execution.

Dell Unity, version(s) 5.4 and prior, contain(s) an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Command execution and Elevation of privileges.

NVIDIA WebDataset for all platforms contains a vulnerability where an attacker could execute arbitrary code with elevated permissions. A successful exploit of this vulnerability might lead to escalation of privileges, data tampering, information disclosure, and denial of service.

SOUND4 IMPACT/FIRST/PULSE/Eco <=2.x contains a command injection vulnerability that allows local authenticated users to create malicious files in the /tmp directory with .dns.pid extension. Unauthenticated attackers can execute the malicious commands by making a single HTTP POST request to the vulnerable dns.php script, which triggers command execution and then deletes the file.

SOUND4 IMPACT/FIRST/PULSE/Eco <=2.x contains a conditional command injection vulnerability that allows local authenticated users to create malicious files in the /tmp directory. Unauthenticated attackers can execute commands by making a single HTTP POST request to the vulnerable ping.php script, which triggers the malicious file and then deletes it.

PraisonAI is a multi-agent teams system. Prior to version 1.5.90, run_python() in praisonai constructs a shell command string by interpolating user-controlled code into python3 -c "<code>" and passing it to subprocess.run(..., shell=True). The escaping logic only handles \ and ", leaving $() and backtick substitutions unescaped, allowing arbitrary OS command execution before Python is invoked. This issue has been patched in version 1.5.90.

Coolify is an open-source and self-hostable tool for managing servers, applications, and databases. Starting in version 4.0.0-beta.18 and prior to 4.0.0-beta.253, a vulnerability in the execution of commands on remote servers allows an authenticated user to execute arbitrary code on the local Coolify container, gaining access to data and private keys or tokens of other users/teams. The ability to inject malicious commands into the Coolify container gives authenticated attackers the ability to fully retrieve and control the data and availability of the software. Centrally hosted Coolify instances (open registration and/or multiple teams with potentially untrustworthy users) are especially at risk, as sensitive data of all users and connected servers can be leaked by any user. Additionally, attackers are able to modify the running software, potentially deploying malicious images to remote nodes or generally changing its behavior. Version 4.0.0-beta.253 patches this issue.

Coolify is an open-source and self-hostable tool for managing servers, applications, and databases. In version 4.0.0-beta.358 and possibly earlier versions, when creating or updating a "project," it is possible to inject arbitrary shell commands by altering the project name. If a name includes unescaped characters, such as single quotes (`'`), it breaks out of the intended command structure, allowing attackers to execute arbitrary commands on the host system. This vulnerability allows attackers to execute arbitrary commands on the host server, which could result in full system compromise; create, modify, or delete sensitive system files; and escalate privileges depending on the permissions of the executed process. Attackers with access to project management features could exploit this flaw to gain unauthorized control over the host environment. Version 4.0.0-beta.359 fixes this issue.

LLama-Index CLI version v0.12.20 contains an OS command injection vulnerability. The vulnerability arises from the improper handling of the `--files` argument, which is directly passed into `os.system`. An attacker who controls the content of this argument can inject and execute arbitrary shell commands. This vulnerability can be exploited locally if the attacker has control over the CLI arguments, and remotely if a web application calls the LLama-Index CLI with a user-controlled filename. This issue can lead to arbitrary code execution on the affected system.

Glances is an open-source system cross-platform monitoring tool. Prior to version 4.5.3, Glances supports dynamic configuration values in which substrings enclosed in backticks are executed as system commands during configuration parsing. This behavior occurs in Config.get_value() and is implemented without validation or restriction of the executed commands. If an attacker can modify or influence configuration files, arbitrary commands will execute automatically with the privileges of the Glances process during startup or configuration reload. In deployments where Glances runs with elevated privileges (e.g., as a system service), this may lead to privilege escalation. This issue has been patched in version 4.5.3.

Multiple local privilege escalation vulnerabilities have been identified in the LiquidVPN client through 1.37 for macOS. An attacker can communicate with an unprotected XPC service and directly execute arbitrary OS commands as root or load a potentially malicious kernel extension because com.smr.liquidvpn.OVPNHelper uses the system function to execute the "openvpncmd" parameter as a shell command.

OpenClaw versions prior to 2026.2.19 contain a local command injection vulnerability in Windows scheduled task script generation due to unsafe handling of cmd metacharacters and expansion-sensitive characters in gateway.cmd files. Local attackers with control over service script generation arguments can inject arbitrary commands by providing metacharacter-only values or CR/LF sequences that execute unintended code in the scheduled task context.

OpenClaw versions 2026.2.26 prior to 2026.3.1 on Windows contain a current working directory injection vulnerability in wrapper resolution for .cmd/.bat files that allows attackers to influence execution behavior through cwd manipulation. Remote attackers can exploit improper shell execution fallback mechanisms to achieve command execution integrity loss by controlling the current working directory during wrapper resolution.

Dell PowerProtect Data Domain with Data Domain Operating System (DD OS) of Feature Release versions 7.7.1.0 through 8.1.0.10, LTS2024 release Versions 7.13.1.0 through 7.13.1.25, LTS 2023 release versions 7.10.1.0 through 7.10.1.50, contain an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability in the DDSH CLI. A low privileged attacker with local access could potentially exploit this vulnerability to execute arbitrary commands with root privileges.

Phicomm K2 v22.6.534.263 was discovered to contain a command injection vulnerability via the autoUpTime parameter in the automatic upgrade function.

The default console presented to users over telnet (when enabled) is restricted to a subset of commands. Commands issued at this console, however, appear to be fed directly into a system call or other similar function. This allows any authenticated user to execute arbitrary commands on the device.

The CLI command parser on Cisco RV110W, RV130W, and RV215W devices allows local users to execute arbitrary shell commands as an administrator via crafted parameters, aka Bug IDs CSCuv90134, CSCux58161, and CSCux73567.

Ericsson Indoor Connect 8855 contains a command injection vulnerability which if exploited can result in an escalation of privileges.

An issue was discovered on Systrome Cumilon ISG-600C, ISG-600H, and ISG-800W devices with firmware V1.1-R2.1_TRUNK-20181105.bin. A shell command injection occurs by editing the description of an ISP file. The file network/isp/isp_update_edit.php does not properly validate user input, which leads to shell command injection via the des parameter.

Dell Unity, version(s) 5.4 and prior, contain(s) an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Code execution and Elevation of privileges.

evernote-mcp-server openBrowser Command Injection Privilege Escalation Vulnerability. This vulnerability allows local attackers to escalate privileges on affected installations of evernote-mcp-server. An attacker must first obtain the ability to execute low-privileged code on the target system in order to exploit this vulnerability. The specific flaw exists within the openBrowser function. The issue results from the lack of proper validation of a user-supplied string before using it to execute a system call. An attacker can leverage this vulnerability to escalate privileges and execute arbitrary code in the context of the service account. Was ZDI-CAN-27913.

Dell Unity, version(s) 5.4 and prior, contain(s) an Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Command execution and Elevation of privileges.

Fleet is open source device management software. Prior to 4.81.1, the Orbit agent's FileVault disk encryption key rotation flow on collects a local user's password via a GUI dialog and interpolates it directly into a Tcl/expect script executed via exec.Command("expect", "-c", script). Because the password is inserted into Tcl brace-quoted send {%s}, a password containing } terminates the literal and injects arbitrary Tcl commands. Since Orbit runs as root, this allows a local unprivileged user to escalate to root privileges. This vulnerability is fixed in 4.81.1.