| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| mudler/LocalAI version 2.17.1 allows for arbitrary file write due to improper handling of automatic archive extraction. When model configurations specify additional files as archives (e.g., .tar), these archives are automatically extracted after downloading. This behavior can be exploited to perform a 'tarslip' attack, allowing files to be written to arbitrary locations on the server, bypassing checks that normally restrict files to the models directory. This vulnerability can lead to remote code execution (RCE) by overwriting backend assets used by the server. |
| A vulnerability in corydolphin/flask-cors version 4.0.1 allows for inconsistent CORS matching due to the handling of the '+' character in URL paths. The request.path is passed through the unquote_plus function, which converts the '+' character to a space ' '. This behavior leads to incorrect path normalization, causing potential mismatches in CORS configuration. As a result, endpoints may not be matched correctly to their CORS settings, leading to unexpected CORS policy application. This can cause unauthorized cross-origin access or block valid requests, creating security vulnerabilities and usability issues. |
| In version 1.5.5 of mintplex-labs/anything-llm, the `/setup-complete` API endpoint allows unauthorized users to access sensitive system settings. The data returned by the `currentSettings` function includes sensitive information such as API keys for search engines, which can be exploited by attackers to steal these keys and cause loss of user assets. |
| BerriAI/litellm version 1.40.12 contains a vulnerability that allows remote code execution. The issue exists in the handling of the 'post_call_rules' configuration, where a callback function can be added. The provided value is split at the final '.' mark, with the last part considered the function name and the remaining part appended with the '.py' extension and imported. This allows an attacker to set a system method, such as 'os.system', as a callback, enabling the execution of arbitrary commands when a chat response is processed. |
| A path traversal vulnerability exists in the `apply_settings` function of parisneo/lollms versions prior to 9.5.1. The `sanitize_path` function does not adequately secure the `discussion_db_name` parameter, allowing attackers to manipulate the path and potentially write to important system folders. |
| A path traversal vulnerability exists in gaizhenbiao/chuanhuchatgpt version 20240410, allowing any user to delete other users' chat histories. This vulnerability can also be exploited to delete any files ending in `.json` on the target system, leading to a denial of service as users are unable to authenticate. |
| An improper access control vulnerability exists in lunary-ai/lunary at the latest commit (a761d83) on the main branch. The vulnerability allows an attacker to use the auth tokens issued by the 'invite user' functionality to obtain valid JWT tokens. These tokens can be used to compromise target users upon registration for their own arbitrary organizations. The attacker can invite a target email, obtain a one-time use token, retract the invite, and later use the token to reset the password of the target user, leading to full account takeover. |
| In version 1.2.7 of lunary-ai/lunary, any authenticated user, regardless of their role, can change the name of an organization due to improper access control. The function checkAccess() is not implemented, allowing users with the lowest privileges, such as the 'Prompt Editor' role, to modify organization attributes without proper authorization. |
| In parisneo/lollms-webui version v9.8, the lollms_binding_infos is missing the client_id parameter, which leads to multiple security vulnerabilities. Specifically, the endpoints /reload_binding, /install_binding, /reinstall_binding, /unInstall_binding, /set_active_binding_settings, and /update_binding_settings are susceptible to CSRF attacks and local attacks. An attacker can exploit this vulnerability to perform unauthorized actions on the victim's machine. |
| A Regular Expression Denial of Service (ReDoS) vulnerability exists in the latest version of gaizhenbiao/chuanhuchatgpt. The vulnerability is located in the filter_history function within the utils.py module. This function takes a user-provided keyword and attempts to match it against chat history filenames using a regular expression search. Due to the lack of sanitization or validation of the keyword parameter, an attacker can inject a specially crafted regular expression, leading to a denial of service condition. This can cause severe degradation of service performance and potential system unavailability. |
| A vulnerability in gaizhenbiao/chuanhuchatgpt version 20240410 allows an attacker to create arbitrary folders at any location on the server, including the root directory (C: dir). This can lead to uncontrolled resource consumption, resulting in resource exhaustion, denial of service (DoS), server unavailability, and potential data loss or corruption. |
| A vulnerability in the /v1/runs API endpoint of lightning-ai/pytorch-lightning v2.2.4 allows attackers to exploit path traversal when extracting tar.gz files. When the LightningApp is running with the plugin_server, attackers can deploy malicious tar.gz plugins that embed arbitrary files with path traversal vulnerabilities. This can result in arbitrary files being written to any directory in the victim's local file system, potentially leading to remote code execution. |
| In h2oai/h2o-3 version 3.46.0, the `run_tool` command in the `rapids` component allows the `main` function of any class under the `water.tools` namespace to be called. One such class, `MojoConvertTool`, crashes the server when invoked with an invalid argument, causing a denial of service. |
| Vanna v0.3.4 is vulnerable to SQL injection in its DuckDB integration exposed to its Flask Web APIs. Attackers can inject malicious SQL training data and generate corresponding queries to write arbitrary files on the victim's file system, such as backdoor.php with contents `<?php system($_GET[0]); ?>`. This can lead to command execution or the creation of backdoors. |
| The vulnerability allows an attacker to access sensitive files on the server by confusing the agent with incorrect file names. When a user requests the content of a file with a misspelled name, the agent attempts to correct the command and inadvertently reveals the content of the intended file, such as /etc/passwd. This can lead to unauthorized access to sensitive information and potential server compromise. |
| vanna-ai/vanna version v0.3.4 is vulnerable to SQL injection in some file-critical functions such as `pg_read_file()`. This vulnerability allows unauthenticated remote users to read arbitrary local files on the victim server, including sensitive files like `/etc/passwd`, by exploiting the exposed SQL queries via a Python Flask API. |
| In lunary-ai/lunary version 1.2.4, an improper access control vulnerability allows members with team management permissions to manipulate project identifiers in requests, enabling them to invite users to projects in other organizations, change members to projects in other organizations with escalated privileges, and change members from other organizations to their own or other projects, also with escalated privileges. This vulnerability is due to the backend's failure to validate project identifiers against the current user's organization ID and projects belonging to it, as well as a misconfiguration in attribute naming (`org_id` should be `orgId`) that prevents proper user organization validation. As a result, attackers can cause inconsistencies on the platform for affected users and organizations, including unauthorized privilege escalation. The issue is present in the backend API endpoints for user invitation and modification, specifically in the handling of project IDs in requests. |
| berriai/litellm version 1.34.34 is vulnerable to improper access control in its team management functionality. This vulnerability allows attackers to perform unauthorized actions such as creating, updating, viewing, deleting, blocking, and unblocking any teams, as well as adding or deleting any member to or from any teams. The vulnerability stems from insufficient access control checks in various team management endpoints, enabling attackers to exploit these functionalities without proper authorization. |
| A Denial of Service (DoS) vulnerability exists in the jaraco/zipp library, affecting all versions prior to 3.19.1. The vulnerability is triggered when processing a specially crafted zip file that leads to an infinite loop. This issue also impacts the zipfile module of CPython, as features from the third-party zipp library are later merged into CPython, and the affected code is identical in both projects. The infinite loop can be initiated through the use of functions affecting the `Path` module in both zipp and zipfile, such as `joinpath`, the overloaded division operator, and `iterdir`. Although the infinite loop is not resource exhaustive, it prevents the application from responding. The vulnerability was addressed in version 3.19.1 of jaraco/zipp. |
| In h2oai/h2o-3 version 3.40.0.4, an exposure of sensitive information vulnerability exists due to an arbitrary system path lookup feature. This vulnerability allows any remote user to view full paths in the entire file system where h2o-3 is hosted. Specifically, the issue resides in the Typeahead API call, which when requested with a typeahead lookup of '/', exposes the root filesystem including directories such as /home, /usr, /bin, among others. This vulnerability could allow attackers to explore the entire filesystem, and when combined with a Local File Inclusion (LFI) vulnerability, could make exploitation of the server trivial. |
