OpenClaw is a personal AI assistant. Prior to version 2026.2.15, OpenClaw embedded the current working directory (workspace path) into the agent system prompt without sanitization. If an attacker can cause OpenClaw to run inside a directory whose name contains control/format characters (for example newlines or Unicode bidi/zero-width markers), those characters could break the prompt structure and inject attacker-controlled instructions. Starting in version 2026.2.15, the workspace path is sanitized before it is embedded into any LLM prompt output, stripping Unicode control/format characters and explicit line/paragraph separators. Workspace path resolution also applies the same sanitization as defense-in-depth.

OpenClaw through 2026.2.22 contains a symlink traversal vulnerability in agents.create and agents.update handlers that use fs.appendFile on IDENTITY.md without symlink containment checks. Attackers with workspace access can plant symlinks to append attacker-controlled content to arbitrary files, enabling remote code execution via crontab injection or unauthorized access via SSH key manipulation.

OpenClaw before 2026.3.25 contains a privilege escalation vulnerability where silent local shared-auth reconnects auto-approve scope-upgrade requests, widening paired device permissions from operator.read to operator.admin. Attackers can exploit this by triggering local reconnection to silently escalate privileges and achieve remote code execution on the node.

OpenClaw versions prior to 2026.2.22 on macOS contain a path validation bypass vulnerability in the exec-approval allowlist mode that allows local attackers to execute unauthorized binaries by exploiting basename-only allowlist entries. Attackers can execute same-name local binaries ./echo without approval when security=allowlist and ask=on-miss are configured, bypassing intended path-based policy restrictions.

OpenClaw versions 2026.1.21 prior to 2026.2.19 contain a path hijacking vulnerability in tools.exec.safeBins that allows attackers to bypass allowlist checks by controlling process PATH resolution. Attackers who can influence the gateway process PATH or launch environment can execute trojan binaries with allowlisted names, such as jq, circumventing executable validation controls.

OpenClaw versions prior to 2026.3.1 fail to properly handle authentication bootstrap errors during startup, allowing browser-control routes to remain accessible without authentication. Local processes or loopback-reachable SSRF paths can exploit this to access browser-control routes including evaluate-capable actions without valid credentials.

OpenClaw versions prior to 2026.2.22 contain an arbitrary shell execution vulnerability in shell environment fallback that trusts the unvalidated SHELL path from the host environment. An attacker with local environment access can inject a malicious SHELL variable to execute arbitrary commands with the privileges of the OpenClaw process.

OpenClaw versions prior to 2026.2.24 contain a policy bypass vulnerability in the safeBins allowlist evaluation that trusts static default directories including writable package-manager paths like /opt/homebrew/bin and /usr/local/bin. An attacker with write access to these trusted directories can place a malicious binary with the same name as an allowed executable to achieve arbitrary command execution within the OpenClaw runtime 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.

OpenClaw versions prior to 2026.3.2 contain a vulnerability in the stageSandboxMedia function in which it fails to validate destination symlinks during media staging, allowing writes to follow symlinks outside the sandbox workspace. Attackers can exploit this by placing symlinks in the media/inbound directory to overwrite arbitrary files on the host system outside sandbox boundaries.

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 prior to 2026.2.25 contain a symlink traversal vulnerability in browser trace and download output path handling that allows local attackers to escape the managed temp root directory. An attacker with local access can create symlinks to route file writes outside the intended temp directory, enabling arbitrary file overwrite on the affected system.

OpenClaw versions prior to 2026.2.19 contain a command injection vulnerability in the Lobster extension tool execution that uses Windows shell fallback with shell: true after spawn failures. Attackers can inject shell metacharacters in command arguments to execute arbitrary commands when subprocess launch fails with EINVAL or ENOENT errors.

OpenClaw versions 2026.1.5 prior to 2026.2.12 fail to enforce mandatory authentication on the /agent/act browser-control HTTP route, allowing unauthorized local callers to invoke privileged operations. Remote attackers on the local network or local processes can execute arbitrary browser-context actions and access sensitive in-session data by sending requests to unauthenticated endpoints.

OpenClaw versions prior to 2026.2.19 contain a command injection vulnerability in Windows Scheduled Task script generation where environment variables are written to gateway.cmd using unquoted set KEY=VALUE assignments, allowing shell metacharacters to break out of assignment context. Attackers can inject arbitrary commands through environment variable values containing metacharacters like &, |, ^, %, or ! to achieve command execution when the scheduled task script is generated and executed.

OpenClaw version 2026.2.22 prior to 2026.2.23 contain an arbitrary code execution vulnerability in shell-env that allows attackers to execute attacker-controlled binaries by exploiting trusted-prefix fallback logic for the $SHELL variable. An attacker can influence the $SHELL environment variable on systems with writable trusted-prefix directories such as /opt/homebrew/bin to execute arbitrary binaries in the OpenClaw process context.

OpenClaw versions prior to 2026.2.24 contain a command injection vulnerability in the system.run shell-wrapper that allows attackers to execute hidden commands by injecting positional argv carriers after inline shell payloads. Attackers can craft misleading approval text while executing arbitrary commands through trailing positional arguments that bypass display context validation.

NFX Series devices using Juniper Networks Junos OS are susceptible to a local command execution vulnerability thereby allowing an attacker to elevate their privileges via the Junos Device Management Daemon (JDMD) process. This issue affects Juniper Networks Junos OS on NFX Series 17.2 version 17.2R1 and later versions prior to 18.3R3-S4; 18.4 versions prior to 18.4R2-S5, 18.4R3-S5; 19.1 versions prior to 19.1R1-S3; 19.2 version 19.1R2 and later versions prior to 19.2R3; 19.3 versions prior to 19.3R3; 19.4 versions prior to 19.4R2-S2. 19.4 versions 19.4R3 and above. This issue does not affect Juniper Networks Junos OS versions prior to 17.2R1. This issue does not affect the JDMD as used by Junos Node Slicing such as External Servers use in conjunction with Junos Node Slicing and In-Chassis Junos Node Slicing on MX480, MX960, MX2008, MX2010, MX2020.

Versions of the package mt7688-wiscan before 0.8.3 are vulnerable to Command Injection due to improper input sanitization in the 'wiscan.scan' function.

All versions of the package puppet-facter are vulnerable to Command Injection via the getFact function due to improper input sanitization.

A vulnerability in the CLI of Cisco SD-WAN Solution software could allow an authenticated, local attacker to inject arbitrary commands that are executed with root privileges. The vulnerability is due to insufficient input validation. An attacker could exploit this vulnerability by authenticating to the device and submitting crafted input to the CLI utility. The attacker must be authenticated to access the CLI utility. A successful exploit could allow the attacker to execute commands with root privileges.

All versions of the package create-choo-app3 are vulnerable to Command Injection via the devInstall function due to improper user-input sanitization.

A vulnerability has been identified in SINEMA Remote Connect Client (All versions < V3.2 HF1). The system service of affected applications is vulnerable to command injection due to missing server side input sanitation when loading VPN configurations. This could allow an authenticated local attacker to execute arbitrary code with system privileges.

Dell SmartFabric OS10 Software, version(s) 10.5.4.x, 10.5.5.x, 10.5.6.x, 10.6.0.x, contain(s) an Improper Neutralization of Special Elements used in a Command ('Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to execution of commands with elevated privileges.

A shell command injection vulnerability in the PAN-OS CLI allows a local authenticated user to escape the restricted shell and escalate privileges. This issue affects only PAN-OS 8.1 versions earlier than PAN-OS 8.1.13. This issue does not affect PAN-OS 7.1, PAN-OS 9.0, or later PAN-OS versions. This issue is fixed in PAN-OS 8.1.13, and all later versions.

A command injection vulnerability in the szc script of the ccurtsinger/stabilizer repository allows remote attackers to execute arbitrary system commands via unsanitized user input passed to os.system(). The vulnerability arises from improper input handling where command-line arguments are directly concatenated into shell commands without validation

A security vulnerability has been detected in tufantunc ssh-mcp up to 1.5.0. The affected element is the function shell.write of the file src/index.ts. Such manipulation of the argument Description leads to command injection. The attack must be carried out locally. The exploit has been disclosed publicly and may be used. The project was informed of the problem early through an issue report but has not responded yet.

A vulnerability has been identified in SINEMA Remote Connect Client (All versions < V3.2 HF1). The system service of affected applications is vulnerable to command injection due to missing server side input sanitation when loading proxy configurations. This could allow an authenticated local attacker to execute arbitrary code with system privileges.

Inspektor Gadget is a set of tools and framework for data collection and system inspection on Kubernetes clusters and Linux hosts using eBPF. The `ig` binary provides a subcommand for image building, used to generate custom gadget OCI images. A part of this functionality is implemented in the file `inspektor-gadget/cmd/common/image/build.go`. The `Makefile.build` file is the Makefile template employed during the building process. This file includes user-controlled data in an unsafe fashion, specifically some parameters are embedded without an adequate escaping in the commands inside the Makefile. Prior to version 0.48.1, this implementation is vulnerable to command injection: an attacker able to control values in the `buildOptions` structure would be able to execute arbitrary commands during the building process. An attacker able to exploit this vulnerability would be able to execute arbitrary command on the Linux host where the `ig` command is launched, if images are built with the `--local` flag or on the build container invoked by `ig`, if the `--local` flag is not provided. The `buildOptions` structure is extracted from the YAML gadget manifest passed to the `ig image build` command. Therefore, the attacker would need a way to control either the full `build.yml` file passed to the `ig image build` command, or one of its options. Typically, this could happen in a CI/CD scenario that builds untrusted gadgets to verify correctness. Version 0.48.1 fixes the issue.

All versions of package git-archive are vulnerable to Command Injection via the exports function.

A vulnerability, which was classified as critical, has been found in exelban stats up to 2.11.21. This issue affects the function shouldAcceptNewConnection of the component XPC Service. The manipulation leads to command injection. It is possible to launch the attack on the local host. Upgrading to version 2.11.22 is able to address this issue. It is recommended to upgrade the affected component.

An issue was discovered in SaltStack Salt before 3002.5. The minion's restartcheck is vulnerable to command injection via a crafted process name. This allows for a local privilege escalation by any user able to create a files on the minion in a non-blacklisted directory.

virtualenv before 20.26.6 allows command injection through the activation scripts for a virtual environment. Magic template strings are not quoted correctly when replacing. NOTE: this is not the same as CVE-2024-9287.

Dell ThinOS 10 versions prior to ThinOS 2602_10.0573, contain an Improper Neutralization of Special Elements used in a Command ('Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Elevation of Privileges.

An attacker with local access the to medical office computer can escalate his Windows user privileges to "NT AUTHORITY\SYSTEM" by exploiting a command injection vulnerability in the Elefant Update Service. The command injection can be exploited by communicating with the Elefant Update Service which is running as "SYSTEM" via Windows Named Pipes.The Elefant Software Updater (ESU) consists of two components. An ESU service which runs as "NT AUTHORITY\SYSTEM" and an ESU tray client which communicates with the service to update or repair the installation and is running with user permissions. The communication is implemented using named pipes. A crafted message of type "MessageType.SupportServiceInfos" can be sent to the local ESU service to inject commands, which are then executed as "NT AUTHORITY\SYSTEM".

Certain NETGEAR devices are affected by command injection by an authenticated user. This affects XR500 before 2.3.2.56 and XR700 before 1.0.1.20.

Dell SmartFabric OS10 Software, version(s) 10.5.6.x, 10.5.5.x, 10.5.4.x, 10.5.3.x, contain(s) an Improper Neutralization of Special Elements used in a Command ('Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Code execution.

Dell SmartFabric OS10 Software, version(s) 10.5.4.x, 10.5.5.x, 10.5.6.x, 10.6.0.x, contain(s) an Improper Neutralization of Special Elements used in a Command ('Command Injection') vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Command execution.

A local privilege escalation vlnerability in the WatchGuard Mobile VPN with SSL client on Windows enables a local user to execute arbitrary commands with elevated privileged.

Dell SmartFabric OS10 Software, version(s) 10.5.6.x, 10.5.5.x, 10.5.4.x, 10.5.3.x, contain(s) a command injection vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Command execution.

An OS command injection vulnerability has been reported to affect several product versions. If exploited, the vulnerability could allow local network attackers to execute commands. We have already fixed the vulnerability in the following versions: QuRouter 2.4.4.106 and later

A vulnerability was determined in neurobin shc up to 4.0.3. This vulnerability affects the function make of the file src/shc.c of the component Filename Handler. Executing manipulation can lead to os command injection. The attack can only be executed locally. The exploit has been publicly disclosed and may be utilized.

A security flaw has been discovered in neurobin shc up to 4.0.3. Impacted is the function make of the file src/shc.c of the component Environment Variable Handler. The manipulation results in os command injection. The attack is only possible with local access. The exploit has been released to the public and may be exploited.

An OS command injection vulnerability has been reported to affect several QNAP operating system versions. If exploited, the vulnerability could allow local network users to execute commands via unspecified vectors. We have already fixed the vulnerability in the following versions: QTS 5.1.8.2823 build 20240712 and later QuTS hero h5.1.8.2823 build 20240712 and later

VMware Aria Operations contains a local privilege escalation vulnerability.  A malicious actor with local administrative privileges can insert malicious commands into the properties file to escalate privileges to  a root user on the appliance running VMware Aria Operations.

A vulnerability classified as critical has been found in Eluktronics Control Center 5.23.51.41. Affected is an unknown function of the file \AiStoneService\MyControlCenter\Command of the component Powershell Script Handler. The manipulation leads to command injection. Attacking locally is a requirement. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.

A command injection is possible through the user interface, allowing arbitrary command execution as the root user. oMG2000 running MGOS 3.15.1 or earlier is affected.  MG90 running MGOS 4.2.1 or earlier is affected.

Cisco Adaptive Security Appliance (ASA) Software before 8.4(1) on ASA 5500, ASA 5500-X, PIX, and FWSM devices allows local users to gain privileges via invalid CLI commands, aka Bug ID CSCtu74257 or EPICBANANA.

An arbitrary file creation vulnerability exists in PaperCut NG/MF that only affects Windows servers with Web Print enabled. This specific flaw exists within the image-handler process, which can incorrectly create files that don’t exist when a maliciously formed payload is provided. This can lead to local privilege escalation. Note: This CVE has been split into two (CVE-2024-4712 and CVE-2024-8405) and it’s been rescored with a "Privileges Required (PR)" rating of low, and “Attack Complexity (AC)” rating of low, reflecting the worst-case scenario where an Administrator has granted local login access to standard network users on the host server.

In Autoswitch Python Virtualenv before version 0.16.0, a user who enters a directory with a malicious `.venv` file could run arbitrary code without any user interaction. This is fixed in version: 1.16.0