LMDeploy is a toolkit for compressing, deploying, and serving large language models. In versions 0.12.3 and prior, hardcoded "trust_remote_code=True" enables HF supply-chain RCE without user opt-in. At time of publication, there are no publicly available patches.

Inkdrop prior to v5.6.0 allows a local attacker to conduct a code injection attack by having a legitimate user open a specially crafted markdown file.

Archive command in Chef InSpec prior to 4.56.58 and 5.22.29 allow local command execution via maliciously crafted profile.

An issue in CMSmadesimple v.2.2.18 allows a local attacker to execute arbitrary code via a crafted payload to the Content Manager Menu component.

claude-code-cache-fix is a cache optimization proxy for Claude Code. From 3.5.0 to before 3.5.2, tools/quota-statusline.sh (introduced in v3.5.0) interpolates Claude Code's hook stdin payload directly into a Python triple-quoted string literal. A ''' byte sequence in any user-controlled field of the payload closes the literal early and lets following bytes execute as Python in the user's Claude Code process. This vulnerability is fixed in 3.5.2.

uniget is a universal installer and updater for (container) tools. Prior to 0.27.1, a command injection vulnerability exists in uniget due to unsafe execution of the check field from metadata files using /bin/bash -c. Because the check field is loaded directly from untrusted JSON metadata without validation or sanitization, an attacker can craft malicious metadata that executes arbitrary shell commands on the victim’s system when common uniget operations such as describe, install, update, or inspect are performed. This vulnerability can lead to arbitrary code execution with the privileges of the user running uniget. This vulnerability is fixed in 0.27.1.

Roslyn CodeLens MCP Server is a Roslyn-based MCP server providing semantic code intelligence for .NET codebases. From 0.0.9 to 1.17.0, the get_diagnostics MCP tool loads and executes all DiagnosticAnalyzer assemblies referenced by the target solution without any allowlist, signature check, or user confirmation; includeAnalyzers defaults to true, so no explicit opt-in is required. An attacker who can place a malicious .csproj referencing an attacker-controlled DLL in a location the victim opens with the MCP server will achieve arbitrary code execution in the server process with the server's OS privileges. This vulnerability is fixed in 1.17.0.

The issue was addressed with improved memory handling. This issue is fixed in macOS Ventura 13.6, tvOS 17, iOS 16.7 and iPadOS 16.7, macOS Monterey 12.7, watchOS 10, iOS 17 and iPadOS 17, macOS Sonoma 14. An app may be able to execute arbitrary code with kernel privileges.

electerm is an open-sourced terminal/ssh/sftp/telnet/serialport/RDP/VNC/Spice/ftp client. Prior to version 3.7.9, a code execution (RCE) vulnerability exists in electerm's SFTP open with system editor or "Edit with custom editor" feature. When a user opts to edit a file using open with system editor or open with a custom editor, the filename is passed directly into a command line without sanitization. A malicious actor controlling the SSH server or user OS can exploit this by crafting a filename containing shell metacharacters. If a victim subsequently attempts to edit this file, the injected commands are executed on their machine with the user's privileges. This could allow the attacker to run arbitrary code, install malware, or move laterally within the network. This issue has been patched in version 3.7.9.

yt-dlp is a youtube-dl fork with additional features and fixes. yt-dlp allows the user to provide shell command lines to be executed at various stages in its download steps through the `--exec` flag. This flag allows output template expansion in its argument, so that metadata values may be used in the shell commands. The metadata fields can be combined with the `%q` conversion, which is intended to quote/escape these values so they can be safely passed to the shell. However, the escaping used for `cmd` (the shell used by Python's `subprocess` on Windows) does not properly escape special characters, which can allow for remote code execution if `--exec` is used directly with maliciously crafted remote data. This vulnerability only impacts `yt-dlp` on Windows, and the vulnerability is present regardless of whether `yt-dlp` is run from `cmd` or from `PowerShell`. Support for output template expansion in `--exec`, along with this vulnerable behavior, was added to `yt-dlp` in version 2021.04.11. yt-dlp version 2023.09.24 fixes this issue by properly escaping each special character. `\n` will be replaced by `\r` as no way of escaping it has been found. It is recommended to upgrade yt-dlp to version 2023.09.24 as soon as possible. Also, always be careful when using --exec, because while this specific vulnerability has been patched, using unvalidated input in shell commands is inherently dangerous. For Windows users who are not able to upgrade: 1. Avoid using any output template expansion in --exec other than {} (filepath). 2. If expansion in --exec is needed, verify the fields you are using do not contain ", | or &. 3. Instead of using --exec, write the info json and load the fields from it instead.

Babel is a compiler for writing next generation JavaScript. From 7.12.0 to before 7.29.4 and 8.0.0-alpha.13, using Babel to compile code that was specifically crafted by an attacker can cause Babel to generate output code that executes arbitrary code. This vulnerability is fixed in 7.29.4 and 8.0.0-alpha.13.

Zed is a code editor. Prior to 0.229.0, Zed's terminal tool permission system can be bypassed by prepending environment variable assignments to allowlisted commands, hijacking program behavior (e.g., PAGER) to execute arbitrary code. This vulnerability is fixed in 0.229.0.

pyp2spec generates working Fedora RPM spec file for Python projects. Prior to version 0.14.1, pyp2spec was writing PyPI package metadata (e.g. the summary field) into the generated spec file without escaping RPM macro directives. When a packager then runs rpmbuild, those directives get evaluated, so a malicious package can execute arbitrary commands on the build machine. This issue has been patched in version 0.14.1.

protobufjs-cli is the command line add-on for protobuf.js. Prior to 1.2.1 and 2.0.2, pbts invoked JSDoc by building a shell command string from input file paths and executing it through child_process.exec. File paths containing shell metacharacters could therefore be interpreted by the shell instead of being passed to JSDoc as plain arguments. This vulnerability is fixed in 1.2.1 and 2.0.2.

Multiple OS command injection vulnerabilities exist in the decompression functionality of GTKWave 3.3.115. A specially crafted wave file can lead to arbitrary command execution. A victim would need to open a malicious file to trigger these vulnerabilities.This vulnerability concerns legacy decompression in `vcd_main`.

IBM Spectrum Protect Client and IBM Storage Protect for Virtual Environments 8.1.0.0 through 8.1.19.0 could allow a local user to execute arbitrary code on the system using a specially crafted file, caused by a DLL hijacking flaw. IBM X-Force ID: 259246.

Kitty is a cross-platform GPU based terminal. In versions prior to 0.47.0, a program able to write bytes to a kitty terminal — a remote SSH peer, a downloaded file viewed with `cat`, a log line, an email body rendered in `less`, an issue body in a TUI, etc. — can cause kitty to execute attacker-supplied Python inside the running kitty process, with the user's full privileges. There is no approval prompt, no remote-control permission requirement, no shell-integration interaction, no clipboard touch, and no editor interaction. Version 0.47.0 fixes the issue.

Multiple OS command injection vulnerabilities exist in the decompression functionality of GTKWave 3.3.115. A specially crafted wave file can lead to arbitrary command execution. A victim would need to open a malicious file to trigger these vulnerabilities.This vulnerability concerns decompression in the `vcd2lxt` utility.

Notepad Next is a cross-platform, reimplementation of Notepad++. Prior to version 0.14, NotepadNext's detectLanguageFromExtension() function interpolates a file's extension directly into a Lua script without sanitization. An attacker can craft a filename whose extension contains Lua code, which executes automatically when the victim opens the file in NotepadNext. Because luaL_openlibs() is called unconditionally, the full os, io, and package libraries are available to the injected code, enabling arbitrary command execution. This issue has been patched in version 0.14.

parseusbs before 1.9 contains an OS command injection vulnerability in parseUSBs.py where LNK file paths are passed unsanitized into an os.popen() shell command, allowing arbitrary command execution via crafted .lnk filenames containing shell metacharacters. An attacker can craft a .lnk filename with embedded shell metacharacters that execute arbitrary commands on the forensic examiner's machine during USB artifact parsing.

parseusbs before 1.9 contains an OS command injection vulnerability where the volume listing path argument (-v flag) is passed unsanitized into an os.popen() shell command with ls, allowing arbitrary command injection via crafted volume path arguments containing shell metacharacters. An attacker can provide a crafted volume path via the -v flag that injects arbitrary commands during volume content enumeration.

UAC (Unix-like Artifacts Collector) before 3.3.0-rc1 contains a command injection vulnerability in the placeholder substitution and command execution pipeline where the _run_command() function passes constructed command strings directly to eval without proper sanitization. Attackers can inject shell metacharacters or command substitutions through attacker-controlled inputs including %line% values from foreach iterators and %user% / %user_home% values derived from system files to achieve arbitrary command execution with the privileges of the UAC process.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, PraisonAI automatically loads a file named tools.py from the current working directory to discover and register custom agent tools. This loading process uses importlib.util.spec_from_file_location and immediately executes module-level code via spec.loader.exec_module() without explicit user consent, validation, or sandboxing. The tools.py file is loaded implicitly, even when it is not referenced in configuration files or explicitly requested by the user. As a result, merely placing a file named tools.py in the working directory is sufficient to trigger code execution. This behavior violates the expected security boundary between user-controlled project files (e.g., YAML configurations) and executable code, as untrusted content in the working directory is treated as trusted and executed automatically. If an attacker can place a malicious tools.py file into a directory where a user or automated system (e.g., CI/CD pipeline) runs praisonai, arbitrary code execution occurs immediately upon startup, before any agent logic begins. This vulnerability is fixed in 4.5.128.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, PraisonAI's AST-based Python sandbox can be bypassed using type.__getattribute__ trampoline, allowing arbitrary code execution when running untrusted agent code. The _execute_code_direct function in praisonaiagents/tools/python_tools.py uses AST filtering to block dangerous Python attributes like __subclasses__, __globals__, and __bases__. However, the filter only checks ast.Attribute nodes, allowing a bypass. The sandbox relies on AST-based filtering of attribute access but fails to account for dynamic attribute resolution via built-in methods such as type.getattribute, resulting in incomplete enforcement of security restrictions. The string '__subclasses__' is an ast.Constant, not an ast.Attribute, so it is never checked against the blocked list. This vulnerability is fixed in 4.5.128.

Composer is a dependency manager for PHP. Versions 1.0 through 2.2.26 and 2.3 through 2.9.5 contain a command injection vulnerability in the Perforce::generateP4Command() method, which constructs shell commands by interpolating user-supplied Perforce connection parameters (port, user, client) without proper escaping. An attacker can inject arbitrary commands through these values in a malicious composer.json declaring a Perforce VCS repository, leading to command execution in the context of the user running Composer, even if Perforce is not installed. VCS repositories are only loaded from the root composer.json or the composer config directory, so this cannot be exploited through composer.json files of packages installed as dependencies. Users are at risk if they run Composer commands on untrusted projects with attacker-supplied composer.json files. This issue has been fixed in Composer 2.2.27 (2.2 LTS) and 2.9.6 (mainline).

radare2 prior to commit bc5a890 contains a command injection vulnerability in the afsv/afsvj command path where crafted ELF binaries can embed malicious r2 command sequences as DWARF DW_TAG_formal_parameter names. Attackers can craft a binary with shell commands in DWARF parameter names that execute when radare2 analyzes the binary with aaa and subsequently runs afsvj, allowing arbitrary shell command execution through the unsanitized parameter interpolation in the pfq command string.

radare2 prior to 6.1.4 contains a command injection vulnerability in the PDB parser's print_gvars() function that allows attackers to execute arbitrary commands by crafting a malicious PDB file with newline characters in symbol names. Attackers can inject arbitrary radare2 commands through unsanitized symbol name interpolation in the flag rename command, which are then executed when a user runs the idp command against the malicious PDB file, enabling arbitrary OS command execution through radare2's shell execution operator.

A Code Injection vulnerability affecting SOLIDWORKS Desktop from Release 2025 through Release 2026 could allow an attacker to execute arbitrary code on the user's machine while opening a specially crafted file.

Electron is a framework for writing cross-platform desktop applications using JavaScript, HTML and CSS. Prior to versions 38.8.6, 39.8.1, 40.8.0, and 41.0.0-beta.8, on macOS, app.moveToApplicationsFolder() used an AppleScript fallback path that did not properly handle certain characters in the application bundle path. Under specific conditions, a crafted launch path could lead to arbitrary AppleScript execution when the user accepted the move-to-Applications prompt. Apps are only affected if they call app.moveToApplicationsFolder(). Apps that do not use this API are not affected. This issue has been patched in versions 38.8.6, 39.8.1, 40.8.0, and 41.0.0-beta.8.

BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Prior to 1.4.38, the cloud deployment path in src/bentoml/_internal/cloud/deployment.py was not included in the fix for CVE-2026-33744. Line 1648 interpolates system_packages directly into a shell command using an f-string without any quoting. The generated script is uploaded to BentoCloud as setup.sh and executed on the cloud build infrastructure during deployment, making this a remote code execution on the CI/CD tier. This vulnerability is fixed in 1.4.38.

BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Prior to 1.4.37, the `docker.system_packages` field in `bentofile.yaml` accepts arbitrary strings that are interpolated directly into Dockerfile `RUN` commands without sanitization. Since `system_packages` is semantically a list of OS package names (data), users do not expect values to be interpreted as shell commands. A malicious `bentofile.yaml` achieves arbitrary command execution during `bentoml containerize` / `docker build`. Version 1.4.37 fixes the issue.

Gematik Authenticator securely authenticates users for login to digital health applications. Starting in version 4.12.0 and prior to version 4.16.0, the Mac OS version of the Authenticator is vulnerable to remote code execution, triggered when victims open a malicious file. Update the gematik Authenticator to version 4.16.0 or greater to receive a patch. There are no known workarounds.

org-babel-execute:latex in ob-latex.el in Org Mode through 9.6.1 for GNU Emacs allows attackers to execute arbitrary commands via a file name or directory name that contains shell metacharacters.

sbt is a build tool for Scala, Java, and others. From version 0.9.5 to before version 1.12.7, on Windows, sbt uses Process("cmd", "/c", ...) to run VCS commands (git, hg, svn). The URI fragment (branch, tag, revision) is user-controlled via the build definition and passed to these commands without validation. Because cmd /c interprets &, |, and ; as command separators, a malicious fragment can execute arbitrary commands. This issue has been patched in version 1.12.7.

Improper sanitization in the invocation of ODA File Converter from FreeCAD 0.19 allows an attacker to inject OS commands via a crafted filename.

Foxit PDF Reader and PDF Editor before 11.1 on macOS allow remote attackers to execute arbitrary code via xfa.host.gotoURL in the XFA API.

The Path Sanity Check script of FreeCAD 0.19 is vulnerable to OS command injection, allowing an attacker to execute arbitrary commands via a crafted FCStd document.

Foxit PDF Reader and PDF Editor before 11.1 on macOS allow remote attackers to execute arbitrary code via app.launchURL in the JavaScript API.

An issue found in Wondershare Technology Co.,Ltd Edraw-max v.12.0.4 allows a remote attacker to execute arbitrary commands via the edraw-max_setup_full5371.exe file.

3D Viewer Remote Code Execution Vulnerability

The shell tool within GitHub Copilot CLI versions prior to and including 0.0.422 can allow arbitrary code execution through crafted bash parameter expansion patterns. An attacker who can influence the commands executed by the agent (e.g., via prompt injection through repository files, MCP server responses, or user instructions) can exploit bash parameter transformation operators to execute hidden commands, bypassing the safety assessment that classifies commands as "read-only." This has been patched in version 0.0.423. The vulnerability stems from how the CLI's shell safety assessment evaluates commands before execution. The safety layer parses and classifies shell commands as either read-only (safe) or write-capable (requires user approval). However, several bash parameter expansion features can embed executable code within arguments to otherwise read-only commands, causing them to appear safe while actually performing arbitrary operations. The specific dangerous patterns are ${var@P}, ${var=value} / ${var:=value}, ${!var}, and nested $(cmd) or <(cmd) inside ${...} expansions. An attacker who can influence command text sent to the shell tool - for example, through prompt injection via malicious repository content (README files, code comments, issue bodies), compromised or malicious MCP server responses, or crafted user instructions containing obfuscated commands - could achieve arbitrary code execution on the user's workstation. This is possible even in permission modes that require user approval for write operations, since the commands can appear to use only read-only utilities to ultimately trigger write operations. Successful exploitation could lead to data exfiltration, file modification, or further system compromise.

Sockeye is an open-source sequence-to-sequence framework for Neural Machine Translation built on PyTorch. Sockeye uses YAML to store model and data configurations on disk. Versions below 2.3.24 use unsafe YAML loading, which can be made to execute arbitrary code embedded in config files. An attacker can add malicious code to the config file of a trained model and attempt to convince users to download and run it. If users run the model, the embedded code will run locally. The issue is fixed in version 2.3.24.

emacsclient-mail.desktop in Emacs 28.1 through 28.2 is vulnerable to shell command injections through a crafted mailto: URI. This is related to lack of compliance with the Desktop Entry Specification. It is fixed in 29.0.90

Microsoft Word Remote Code Execution Vulnerability

Obsidian Dataview through 0.4.12-hotfix1 allows eval injection. The evalInContext function in executes user input, which allows an attacker to craft malicious Markdown files that will execute arbitrary code once opened. NOTE: 0.4.13 provides a mitigation for some use cases.

Microsoft Defender Remote Code Execution Vulnerability

Microsoft Excel Remote Code Execution Vulnerability

InfCode's terminal auto-execution module contains a critical command filtering vulnerability that renders its blacklist security mechanism completely ineffective. The predefined blocklist fails to cover native high-risk commands in Windows PowerShell (such as powershell), and the matching algorithm lacks dynamic semantic parsing unable to recognize string concatenation, variable assignment, or double-quote interpolation in Shell syntax. Malicious commands can bypass interception through simple syntax obfuscation. An attacker can construct a file containing malicious instructions for remote code injection. When a user imports and views such a file in the IDE, the Agent executes dangerous PowerShell commands outside the blacklist without user confirmation, resulting in arbitrary command execution or sensitive data leakage.

@asyncapi/java-spring-cloud-stream-template generates a Spring Cloud Stream (SCSt) microservice. In versions prior to 0.7.0 arbitrary code injection was possible when an attacker controls the AsyncAPI document. An example is provided in GHSA-xj6r-2jpm-qvxp. There are no mitigations available and all users are advised to update.

Home assistant is an open source home automation. The Home Assistant Companion for Android app up to version 2023.8.2 is vulnerable to arbitrary URL loading in a WebView. This enables all sorts of attacks, including arbitrary JavaScript execution, limited native code execution, and credential theft. This issue has been patched in version 2023.9.2 and all users are advised to upgrade. There are no known workarounds for this vulnerability. This issue is also tracked as GitHub Security Lab (GHSL) Vulnerability Report: `GHSL-2023-142`.