Exploit Intelligence Platform

MindsDB is a platform for building artificial intelligence from enterprise data. Prior to version 25.11.1, an unauthenticated path traversal in the file upload API lets any caller read arbitrary files from the server filesystem and move them into MindsDB’s storage, exposing sensitive data. The PUT handler in file.py directly joins user-controlled data into a filesystem path when the request body is JSON and source_type is not "url". Only multipart uploads and URL-sourced uploads receive sanitization; JSON uploads lack any call to clear_filename or equivalent checks. This vulnerability is fixed in 25.11.1.

LangChain is a framework for building agents and LLM-powered applications. Prior to versions 0.3.81 and 1.2.5, a serialization injection vulnerability exists in LangChain's dumps() and dumpd() functions. The functions do not escape dictionaries with 'lc' keys when serializing free-form dictionaries. The 'lc' key is used internally by LangChain to mark serialized objects. When user-controlled data contains this key structure, it is treated as a legitimate LangChain object during deserialization rather than plain user data. This issue has been patched in versions 0.3.81 and 1.2.5.

Indico is an event management system that uses Flask-Multipass, a multi-backend authentication system for Flask. Starting in version 2.2 and prior to version 3.3.7, an endpoint used to display details of users listed in certain fields (such as ACLs) could be misused to dump basic user details (such as name, affiliation and email) in bulk. Version 3.3.7 fixes the issue. Owners of instances that allow everyone to create a user account, who wish to truly restrict access to these user details, should consider restricting user search to managers. As a workaround, it is possible to restrict access to the affected endpoints (e.g. in the webserver config), but doing so would break certain form fields which could no longer show the details of the users listed in those fields, so upgrading instead is highly recommended.

NiceGUI is a Python-based UI framework. Versions 3.3.1 and below are subject to a XSS vulnerability through the ui.interactive_image component of NiceGUI. The component renders SVG content using Vue's v-html directive without any sanitization. This allows attackers to inject malicious HTML or JavaScript via the SVG <foreignObject> tag whenever the image component is rendered or updated. This is particularly dangerous for dashboards or multi-user applications displaying user-generated content or annotations. This issue is fixed in version 3.4.0.

A Stored Cross-Site Scripting (XSS) vulnerability in Calibre-Web v0.6.25 allows attackers to inject malicious JavaScript into the 'username' field during user creation. The payload is stored unsanitized and later executed when the /ajax/listusers endpoint is accessed.

An issue was discovered in Django 4.2 before 4.2.24, 5.1 before 5.1.12, and 5.2 before 5.2.6. FilteredRelation is subject to SQL injection in column aliases, using a suitably crafted dictionary, with dictionary expansion, as the **kwargs passed QuerySet.annotate() or QuerySet.alias().

BentoML is a Python library for building online serving systems optimized for AI apps and model inference. In versions 1.4.0 until 1.4.19, the file upload processing system contains an SSRF vulnerability that allows unauthenticated remote attackers to force the server to make arbitrary HTTP requests. The vulnerability stems from the multipart form data and JSON request handlers, which automatically download files from user-provided URLs without validating whether those URLs point to internal network addresses, cloud metadata endpoints, or other restricted resources. The documentation explicitly promotes this URL-based file upload feature, making it an intended design that exposes all deployed services to SSRF attacks by default. Version 1.4.19 contains a patch for the issue.

An issue was discovered in Overhang.IO (tutor-open-edx) (overhangio/tutor) 20.0.2 allowing local unauthorized attackers to gain access to sensitive information due to the absence of proper cache-control HTTP headers and client-side session checks.

python-socketio is a Python implementation of the Socket.IO realtime client and server. A remote code execution vulnerability in python-socketio versions prior to 5.14.0 allows attackers to execute arbitrary Python code through malicious pickle deserialization in multi-server deployments on which the attacker previously gained access to the message queue that the servers use for internal communications. When Socket.IO servers are configured to use a message queue backend such as Redis for inter-server communication, messages sent between the servers are encoded using the `pickle` Python module. When a server receives one of these messages through the message queue, it assumes it is trusted and immediately deserializes it. The vulnerability stems from deserialization of messages using Python's `pickle.loads()` function. Having previously obtained access to the message queue, the attacker can send a python-socketio server a crafted pickle payload that executes arbitrary code during deserialization via Python's `__reduce__` method. This vulnerability only affects deployments with a compromised message queue. The attack can lead to the attacker executing random code in the context of, and with the privileges of a Socket.IO server process. Single-server systems that do not use a message queue, and multi-server systems with a secure message queue are not vulnerable. In addition to making sure standard security practices are followed in the deployment of the message queue, users of the python-socketio package can upgrade to version 5.14.0 or newer, which remove the `pickle` module and use the much safer JSON encoding for inter-server messaging.

pgAdmin versions up to 9.9 are affected by a Remote Code Execution (RCE) vulnerability that occurs when running in server mode and performing restores from PLAIN-format dump files. This issue allows attackers to inject and execute arbitrary commands on the server hosting pgAdmin, posing a critical risk to the integrity and security of the database management system and underlying data.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. In versions 0.6.34 and below, the functionality that inserts custom prompts into the chat window is vulnerable to DOM XSS when 'Insert Prompt as Rich Text' is enabled, since the prompt body is assigned to the DOM sink .innerHtml without sanitisation. Any user with permissions to create prompts can abuse this to plant a payload that could be triggered by other users if they run the corresponding / command to insert the prompt. This issue is fixed in version 0.6.35.

An issue was discovered in 5.1 before 5.1.14, 4.2 before 4.2.26, and 5.2 before 5.2.8. NFKC normalization in Python is slow on Windows. As a consequence, `django.http.HttpResponseRedirect`, `django.http.HttpResponsePermanentRedirect`, and the shortcut `django.shortcuts.redirect` were subject to a potential denial-of-service attack via certain inputs with a very large number of Unicode characters. Earlier, unsupported Django series (such as 5.0.x, 4.1.x, and 3.2.x) were not evaluated and may also be affected. Django would like to thank Seokchan Yoon for reporting this issue.

Starlette is a lightweight ASGI framework/toolkit. Starting in version 0.39.0 and prior to version 0.49.1 , an unauthenticated attacker can send a crafted HTTP Range header that triggers quadratic-time processing in Starlette's FileResponse Range parsing/merging logic. This enables CPU exhaustion per request, causing denial‑of‑service for endpoints serving files (e.g., StaticFiles or any use of FileResponse). This vulnerability is fixed in 0.49.1.

The Keras Model.load_model function permits arbitrary code execution, even with safe_mode=True, through a manually constructed, malicious .keras archive. By altering the config.json file within the archive, an attacker can specify arbitrary Python modules and functions, along with their arguments, to be loaded and executed during model loading.

Deserialization of untrusted data in python in pyfory versions 0.12.0 through 0.12.2, or the legacy pyfury versions from 0.1.0 through 0.10.3: allows arbitrary code execution. An application is vulnerable if it reads pyfory serialized data from untrusted sources. An attacker can craft a data stream that selects pickle-fallback serializer during deserialization, leading to the execution of `pickle.loads`, which is vulnerable to remote code execution. Users are recommended to upgrade to pyfory version 0.12.3 or later, which has removed pickle fallback serializer and thus fixes this issue.

feiskyer mcp-kubernetes-server through 0.1.11 does not consider chained commands in the implementation of --disable-write and --disable-delete, e.g., it allows a "kubectl version; kubectl delete pod" command because the first word (i.e., "version") is not a write or delete operation.

A remote code execution (RCE) vulnerability exists in the ms-swift project version 3.3.0 due to unsafe deserialization in tests/run.py using yaml.load() from the PyYAML library (versions = 5.3.1). If an attacker can control the content of the YAML configuration file passed to the --run_config parameter, arbitrary code can be executed during deserialization. This can lead to full system compromise. The vulnerability is triggered when a malicious YAML file is loaded, allowing the execution of arbitrary Python commands such as os.system(). It is recommended to upgrade PyYAML to version 5.4 or higher, and to use yaml.safe_load() to mitigate the issue.

Copyparty is a portable file server. In versions 1.18.6 and below, when accessing the recent uploads page at `/?ru`, users can filter the results using an input field at the top. This field appends a filter parameter to the URL, which reflects its value directly into a `<script>` block without proper escaping, allowing for reflected Cross-Site Scripting (XSS) and can be exploited against both authenticated and unauthenticated users. This is fixed in version 1.18.7.

A cross-site scripting (XSS) vulnerability in the component /blog/blogpost/add of Mezzanine CMS v6.1.0 allows attackers to execute arbitrary web scripts or HTML via injecting a crafted payload into a blog post.

Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability in Calibre Web, Autocaliweb allows Blind OS Command Injection.This issue affects Calibre Web: 0.6.24 (Nicolette); Autocaliweb: from 0.7.0 before 0.7.1.