Vulnerabilities

In lunary-ai/lunary version 1.2.2, an account hijacking vulnerability exists due to a password reset token leak. A user with a 'viewer' role can exploit this vulnerability to hijack another user's account by obtaining the password reset token. The vulnerability is triggered when the 'viewer' role user sends a specific request to the server, which responds with a password reset token in the 'recoveryToken' parameter. This token can then be used to reset the password of another user's account without authorization. The issue results from an excessive attack surface, allowing lower-privileged users to escalate their privileges and take over accounts.

lunary-ai/lunary version 1.9.34 is vulnerable to an account takeover due to improper authentication in the Google OAuth integration. The application fails to verify the 'aud' (audience) field in the access token issued by Google, which is crucial for ensuring the token is intended for the application. This oversight allows attackers to use tokens issued to malicious applications to gain unauthorized access to user accounts. The issue is resolved in version 1.9.35.

A critical stored Cross-Site Scripting (XSS) vulnerability exists in the Analytics component of lunary-ai/lunary versions up to 1.9.23, where the NEXT_PUBLIC_CUSTOM_SCRIPT environment variable is directly injected into the DOM using dangerouslySetInnerHTML without any sanitization or validation. This allows arbitrary JavaScript execution in all users' browsers if an attacker can control the environment variable during deployment or through server compromise. The vulnerability can lead to complete account takeover, data exfiltration, malware distribution, and persistent attacks affecting all users until the environment variable is cleaned. The issue is fixed in version 1.9.25.

lunary-ai/lunary versions prior to 1.9.24 are vulnerable to stored cross-site scripting (XSS). An unauthenticated attacker can inject malicious JavaScript into the `v1/runs/ingest` endpoint by adding an empty `citations` field, triggering a code path where `dangerouslySetInnerHTML` is used to render attacker-controlled text. This vulnerability allows the execution of arbitrary JavaScript in the context of the user's browser, potentially leading to session hijacking, data theft, or other malicious actions.

In lunary-ai/lunary version v1.4.28, the /bigquery API route lacks proper access control, allowing any logged-in user to create a Datastream to Google BigQuery and export the entire database. This includes sensitive data such as password hashes and secret API keys. The route is protected by a config check (`config.DATA_WAREHOUSE_EXPORTS_ALLOWED`), but it does not verify the user's access level or implement any access control middleware. This vulnerability can lead to the extraction of sensitive data, disruption of services, credential compromise, and service integrity breaches.

In lunary-ai/lunary versions up to and including 1.2.5, an information disclosure vulnerability exists where account recovery hashes of users are inadvertently exposed to unauthorized actors. This issue occurs when authenticated users inspect responses from `GET /v1/users/me` and `GET /v1/users/me/org` endpoints. The exposed account recovery hashes, while not directly related to user passwords, represent sensitive information that should not be accessible to unauthorized parties. Exposing these hashes could potentially facilitate account recovery attacks or other malicious activities. The vulnerability was addressed in version 1.2.6.

A SQL injection vulnerability exists in the `/api/v1/external-users` route of lunary-ai/lunary version v1.4.2. The `order by` clause of the SQL query uses `sql.unsafe` without prior sanitization, allowing for SQL injection. The `orderByClause` variable is constructed without server-side validation or sanitization, enabling an attacker to execute arbitrary SQL commands. Successful exploitation can lead to complete data loss, modification, or corruption.

An improper access control vulnerability in lunary-ai/lunary version 1.3.2 allows an attacker to update the SAML configuration without authorization. This vulnerability can lead to manipulation of authentication processes, fraudulent login requests, and theft of user information. Appropriate access controls should be implemented to ensure that the SAML configuration can only be updated by authorized users.

A Cross-Site Request Forgery (CSRF) vulnerability exists in lunary-ai/lunary version 1.2.34 due to overly permissive CORS settings. This vulnerability allows an attacker to sign up for and create projects or use the instance as if they were a user with local access. The main attack vector is for instances hosted locally on personal machines, which are not publicly accessible. The CORS settings in the backend permit all origins, exposing unauthenticated endpoints to CSRF attacks.

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.

In lunary-ai/lunary version 1.2.13, an insufficient granularity of access control vulnerability allows users to create, update, get, and delete prompt variations for datasets not owned by their organization. This issue arises due to the application not properly validating the ownership of dataset prompts and their variations against the organization or project of the requesting user. As a result, unauthorized modifications to dataset prompts can occur, leading to altered or removed dataset prompts without proper authorization. This vulnerability impacts the integrity and consistency of dataset information, potentially affecting the results of experiments.

In lunary-ai/lunary version v1.2.13, an incorrect authorization vulnerability exists that allows unauthorized users to access and manipulate projects within an organization they should not have access to. Specifically, the vulnerability is located in the `checkProjectAccess` method within the authorization middleware, which fails to adequately verify if a user has the correct permissions to access a specific project. Instead, it only checks if the user is part of the organization owning the project, overlooking the necessary check against the `account_project` table for explicit project access rights. This flaw enables attackers to gain complete control over all resources within a project, including the ability to create, update, read, and delete any resource, compromising the privacy and security of sensitive information.

A Server-Side Request Forgery (SSRF) vulnerability exists in the lunary-ai/lunary application, specifically within the endpoint '/auth/saml/tto/download-idp-xml'. The vulnerability arises due to the application's failure to validate user-supplied URLs before using them in server-side requests. An attacker can exploit this vulnerability by sending a specially crafted request to the affected endpoint, allowing them to make unauthorized requests to internal or external resources. This could lead to the disclosure of sensitive information, service disruption, or further attacks against the network infrastructure. The issue affects the latest version of the application as of the report.