🔍 Search

CVE-2025-27614

Gitk is a Tcl/Tk based Git history browser. Starting with 2.41.0, a Git repository can be crafted in such a way that with some social engineering a user who has cloned the repository can be tricked into running any script (e.g., Bourne shell, Perl, python, ...) supplied by the attacker by invoking gitk filename, where filename has a particular structure. The script is run with the privileges of the user. This vulnerability is fixed in 2.43.7, 2.44.4, 2.45.4, 2.46.4, 2.47.3, 2.48.2, 2.49.1, and 2.50.

CVE-2026-0863

Using string formatting and exception handling, an attacker may bypass n8n's python-task-executor sandbox restrictions and run arbitrary unrestricted python code in the underlying operating system. The vulnerability can be exploited via the Code block by an authenticated user with basic permissions and can lead to a full n8n instance takeover on instances operating under "Internal" execution mode. If the instance is operating under the "External" execution mode (ex. n8n's official Docker image) - arbitrary code execution occurs inside a Sidecar container and not the main node, which significantly reduces the vulnerability impact.

CVE-2026-40287

PraisonAI is a multi-agent teams system. Versions 4.5.138 and below are vulnerable to arbitrary code execution through automatic, unsanitized import of a tools.py file from the current working directory. Components including call.py (import_tools_from_file()), tool_resolver.py (_load_local_tools()), and CLI tool-loading paths blindly import ./tools.py at startup without any validation, sandboxing, or user confirmation. An attacker who can place a malicious tools.py in the directory where PraisonAI is launched (such as through a shared project, cloned repository, or writable workspace) achieves immediate arbitrary python code execution in the host environment. This compromises the full PraisonAI process, the host system, and any connected data or credentials. This issue has been fixed in version 4.5.139.

CVE-2025-70560

Boltz 2.0.0 contains an insecure deserialization vulnerability in its molecule loading functionality. The application uses python pickle to deserialize molecule data files without validation. An attacker with the ability to place a malicious pickle file in a directory processed by boltz can achieve arbitrary code execution when the file is loaded.

CVE-2023-49565

The cbis_manager Podman container is vulnerable to remote command execution via the /api/plugins endpoint. Improper sanitization of the HTTP Headers X-FILENAME, X-PAGE, and X-FIELD allows for command injection. These headers are directly utilized within the subprocess.Popen python function without adequate validation, enabling a remote attacker to execute arbitrary commands on the underlying system by crafting malicious header values within an HTTP request to the affected endpoint. The web service executes with root privileges within the container environment, the demonstrated remote code execution permits an attacker to acquire elevated privileges for the command execution. Restricting access to the management network with an external firewall can partially mitigate this risk.

CVE-2025-54886

skops is a python library which helps users share and ship their scikit-learn based models. In versions 0.12.0 and below, the Card.get_model does not contain any logic to prevent arbitrary code execution. The Card.get_model function supports both joblib and skops for model loading. When loading .skops models, it uses skops' secure loading with trusted type validation, raising errors for untrusted types unless explicitly allowed. However, when non-.zip file formats are provided, the function silently falls back to joblib without warning. Unlike skops, joblib allows arbitrary code execution during loading, bypassing security measures and potentially enabling malicious code execution. This issue is fixed in version 0.13.0.

CVE-2024-6982

A remote code execution vulnerability exists in the Calculate function of parisneo/lollms version 9.8. The vulnerability arises from the use of python's `eval()` function to evaluate mathematical expressions within a python sandbox that disables `__builtins__` and only allows functions from the `math` module. This sandbox can be bypassed by loading the `os` module using the `_frozen_importlib.BuiltinImporter` class, allowing an attacker to execute arbitrary commands on the server. The issue is fixed in version 9.10.

CVE-2025-24359

ASTEVAL is an evaluator of python expressions and statements. Prior to version 1.0.6, if an attacker can control the input to the `asteval` library, they can bypass asteval's restrictions and execute arbitrary python code in the context of the application using the library. The vulnerability is rooted in how `asteval` performs handling of `FormattedValue` AST nodes. In particular, the `on_formattedvalue` value uses the dangerous format method of the str class. The code allows an attacker to manipulate the value of the string used in the dangerous call `fmt.format(__fstring__=val)`. This vulnerability can be exploited to access protected attributes by intentionally triggering an `AttributeError` exception. The attacker can then catch the exception and use its `obj` attribute to gain arbitrary access to sensitive or protected object properties. Version 1.0.6 fixes this issue.

CVE-2023-41334

Astropy is a project for astronomy in python that fosters interoperability between python astronomy packages. Version 5.3.2 of the Astropy core package is vulnerable to remote code execution due to improper input validation in the `TranformGraph().to_dot_graph` function. A malicious user can provide a command or a script file as a value to the `savelayout` argument, which will be placed as the first value in a list of arguments passed to `subprocess.Popen`. Although an error will be raised, the command or script will be executed successfully. Version 5.3.3 fixes this issue.

CVE-2021-24105

Depending on configuration of various package managers it is possible for an attacker to insert a malicious package into a package manager's repository which can be retrieved and used during development, build, and release processes. This insertion could lead to remote code execution. We believe this vulnerability affects multiple package managers across multiple languages, including but not limited to: python/pip, .NET/NuGet, Java/Maven, JavaScript/npm.

Attack scenarios

An attacker could take advantage of this ecosystem-wide issue to cause harm in a variety of ways. The original attack scenarios were discovered by Alex Birsan and are detailed in their whitepaper, Dependency Confusion: How I Hacked Into Apple, Microsoft and Dozens of Other Companies.

• With basic knowledge of the target ecosystems, an attacker could create an empty shell for a package and insert malicious code in the install scripts, give it a high version, and publish it to the public repository. Vulnerable victim machines will download the higher version of the package between the public and private repositories and attempt to install it. Due to code incompatibility it will probably error out upon import or upon compilation, making it easier to detect; however the attacker would have gained code execution by that point.

• An advanced attacker with some inside knowledge of the target could take a copy of a working package, insert the malicious code (in the package itself or in the install), and then publish it to a public repository. The package will likely install and import correctly, granting the attacker an initial foothold and persistence.

These two methods could affect target organizations at any of these various levels:

• Developer machines
• An entire team if the configuration to import the malicious package is uploaded to a code repository
• Continuous integration pipelines if they pull the malicious packages during the build, test, and/or deploy stages
• Customers, download servers, production services if the malicious code has not been detected

This remote code execution vulnerability can only be addressed by reconfiguring installation tools and workflows, and not by correcting anything in the package repositories themselves. See the FAQ section of this CVE for configuration guidance.

CVE-2026-45369

python-utcp is the python implementation of UTCP. Prior to 1.1.3, the _substitute_utcp_args method in cli_communication_protocol.py inserts user-controlled tool_args values directly into shell command strings without any sanitization or escaping. These commands are then executed via /bin/bash -c (Unix) or powershell.exe -Command (Windows), allowing an attacker to inject arbitrary shell commands. This vulnerability is fixed in 1.1.3.

CVE-2026-42313

pyLoad is a free and open-source download manager written in python. Prior to 0.5.0b3.dev100, the set_config_value() API method (@permission(Perms.SETTINGS)) in src/pyload/core/api/__init__.py gates security-sensitive options behind a hand-maintained allowlist ADMIN_ONLY_CORE_OPTIONS. The allowlist contains ("proxy", "username") and ("proxy", "password") — which protect the proxy credentials — but it does not include ("proxy", "enabled"), ("proxy", "host"), ("proxy", "port"), or ("proxy", "type"). Any authenticated user with the non-admin SETTINGS permission can enable proxying and point pyload at any host they control. From that point, every outbound download, captcha fetch, update check, and plugin HTTP call is transparently routed through the attacker. This is a direct continuation of the fix family CVE-2026-33509 / CVE-2026-35463 / CVE-2026-35464 / CVE-2026-35586, each of which patched a different missed option in the same allowlist. This vulnerability is fixed in 0.5.0b3.dev100.

CVE-2026-29075

Mesa is an open-source python library for agent-based modeling, simulating complex systems and exploring emergent behaviors. In version 3.5.0 and prior, checking out of untrusted code in benchmarks.yml workflow may lead to code execution in privileged runner. This issue has been patched via commit c35b8cd.

CVE-2024-41671

Twisted is an event-based framework for internet applications, supporting python 3.6+. The HTTP 1.0 and 1.1 server provided by twisted.web could process pipelined HTTP requests out-of-order, possibly resulting in information disclosure. This vulnerability is fixed in 24.7.0rc1.

CVE-2026-44843

LangChain is a framework for building agents and LLM-powered applications. Prior to 0.3.85 and 1.3.3, LangChain contains older runtime code paths that deserialize run inputs, run outputs, or other application-controlled payloads using overly broad object allowlists. These paths may call load() with allowed_objects="all". This does not enable arbitrary python object deserialization, but it does allow any trusted LangChain-serializable object to be revived, which is broader than these runtime paths require. As a result, attacker-supplied LangChain serialized constructor dictionaries may cause trusted runtime paths to instantiate classes with untrusted constructor arguments. This vulnerability is fixed in 0.3.85 and 1.3.3.

CVE-2026-5843

The MLX inference backend in Docker Model Runner on macOS uses the MLX-LM library, which unconditionally imports and executes arbitrary python files from model directories via the model_file configuration field in config.json. When a model's config.json specifies a model_file pointing to a python file, MLX-LM uses importlib to load and execute it with no trust_remote_code gate or equivalent safety check. The MLX backend runs without sandboxing, resulting in arbitrary code execution on the Docker host as the Docker Desktop user. Any container on the Docker network can trigger this by calling the model-runner.docker.internal API to pull a malicious model from an attacker-controlled OCI registry and request inference.

CVE-2026-5817

The vllm-metal inference backend in Docker Model Runner on macOS unconditionally sets trust_remote_code=True when loading model tokenizers, and runs without sandboxing. This causes transformers.AutoTokenizer.from_pretrained() to import and execute arbitrary python files included in any model pulled from an OCI registry, resulting in arbitrary code execution on the Docker host as the Docker Desktop user when inference is triggered. Any container on the Docker network can trigger this by calling the model-runner.docker.internal API to pull a malicious model and request inference.

CVE-2026-29872

A cross-session information disclosure vulnerability exists in the awesome-llm-apps project in commit e46690f99c3f08be80a9877fab52acacf7ab8251 (2026-01-19). The affected Streamlit-based GitHub MCP Agent stores user-supplied API tokens in process-wide environment variables using os.environ without proper session isolation. Because Streamlit serves multiple concurrent users from a single python process, credentials provided by one user remain accessible to subsequent unauthenticated users. An attacker can exploit this issue to retrieve sensitive information such as GitHub Personal Access Tokens or LLM API keys, potentially leading to unauthorized access to private resources and financial abuse.

CVE-2026-28416

Gradio is an open-source python package designed for quick prototyping. Prior to version 6.6.0, a Server-Side Request Forgery (SSRF) vulnerability in Gradio allows an attacker to make arbitrary HTTP requests from a victim's server by hosting a malicious Gradio Space. When a victim application uses `gr.load()` to load an attacker-controlled Space, the malicious `proxy_url` from the config is trusted and added to the allowlist, enabling the attacker to access internal services, cloud metadata endpoints, and private networks through the victim's infrastructure. Version 6.6.0 fixes the issue.

CVE-2024-53865

zhmcclient is a pure python client library for the IBM Z HMC Web Services API. In affected versions the python package "zhmcclient" writes password-like properties in clear text into its HMC and API logs in the following cases: 1. The 'boot-ftp-password' and 'ssc-master-pw' properties when creating or updating a partition in DPM mode, in the zhmcclient API and HMC logs. 2. The 'ssc-master-pw' and 'zaware-master-pw' properties when updating an LPAR in classic mode, in the zhmcclient API and HMC logs. 3. The 'ssc-master-pw' and 'zaware-master-pw' properties when creating or updating an image activation profile in classic mode, in the zhmcclient API and HMC logs. 4. The 'password' property when creating or updating an HMC user, in the zhmcclient API log. 5. The 'bind-password' property when creating or updating an LDAP server definition, in the zhmcclient API and HMC logs. This issue affects only users of the zhmcclient package that have enabled the python loggers named "zhmcclient.api" (for the API log) or "zhmcclient.hmc" (for the HMC log) and that use the functions listed above. This issue has been fixed in zhmcclient version 1.18.1. Users are advised to upgrade. There are no known workarounds for this vulnerability.