Vulnerabilities

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. Prior to 1.13.0, the filesystem-search-files agent skill passes its LLM-controlled pattern parameter to ripgrep as a positional argument without a -- end-of-options separator. ripgrep parses any argument that starts with - as an option, so a pattern of --pre=/bin/sh turns ripgrep into a script executor: it runs /bin/sh for every file it walks. An attacker who can chat with an agent on a deployment with the filesystem plugin enabled (the default in the official Docker image) can use this, together with the sibling filesystem-write-text-file skill, to run arbitrary commands inside the AnythingLLM server container. This vulnerability is fixed in 1.13.0.

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. Prior to version 1.12.1, GET /api/workspace/:slug/tts/:chatId in AnythingLLM returns the text-to-speech audio for another user's chat response within the same workspace because the route validates workspace membership but does not enforce ownership of the targeted chat row. As a result, an authenticated user can access another user's private assistant response in audio form if the chatId is known or guessed. This constitutes an insecure direct object reference (IDOR) affecting private chat response content exposed through the TTS endpoint. This issue has been patched in version 1.12.1.

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. Prior to version 1.12.1, AnythingLLM's in-chat markdown renderer has an unsafe custom rule for images that interpolates the markdown image's `alt` text into an HTML `alt="..."` attribute without any HTML encoding. Every call-site in the app wraps `renderMarkdown(...)` with `DOMPurify.sanitize(...)` as defense-in-depth — except the `Chartable` component, which renders chart captions with no sanitization. The chart caption is the natural-language text the LLM emits around a `create-chart` tool call, so any attacker who can influence the LLM's output — most cheaply via indirect prompt injection in a shared workspace document, or directly if they can create a chart record in a multi-user workspace — can trigger stored DOM-level XSS in every other user's browser when they open that conversation. AnythingLLM chat history is loaded server-side via `GET /api/workspace/:slug/chats` and rendered directly into the chat UI. Version 1.12.1 contains a patch for this issue.

A path traversal vulnerability exists in mintplex-labs/anything-llm versions up to and including 1.9.1, within the `AgentFlows` component. The vulnerability arises from improper handling of user input in the `loadFlow` and `deleteFlow` methods in `server/utils/agentFlows/index.js`. Specifically, the combination of `path.join` and `normalizePath` allows attackers to bypass directory restrictions and access or delete arbitrary `.json` files on the server. This can lead to information disclosure, such as leaking sensitive configuration files containing API keys, or denial of service by deleting critical files like `package.json`. The issue is resolved in version 1.12.1.

Directory Traversal vulnerability in the extractFile method of org.codehaus.plexus.util.Expand in plexus-utils before 6d780b3378829318ba5c2d29547e0012d5b29642. This allows an attacker to execute arbitrary code

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. In 1.11.1 and earlier, AnythingLLM Desktop contains a Streaming Phase XSS vulnerability in the chat rendering pipeline that escalates to Remote Code Execution on the host OS due to insecure Electron configuration. This works with default settings and requires no user interaction beyond normal chat usage. The custom markdown-it image renderer in frontend/src/utils/chat/markdown.js interpolates token.content directly into the alt attribute without HTML entity escaping. The PromptReply component renders this output via dangerouslySetInnerHTML without DOMPurify sanitization — unlike HistoricalMessage which correctly applies DOMPurify.sanitize().

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. If AnythingLLM prior to version 1.10.0 is configured to use Qdrant as the vector database with an API key, this QdrantApiKey could be exposed in plain text to unauthenticated users via the `/api/setup-complete` endpoint. Leakage of QdrantApiKey allows an unauthenticated attacker full read/write access to the Qdrant vector database instance used by AnythingLLM. Since Qdrant often stores the core knowledge base for RAG in AnythingLLM, this can lead to complete compromise of the semantic search / retrieval functionality and indirect leakage of confidential uploaded documents. Version 1.10.0 patches the issue.

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. Prior to commit e287fab56089cf8fcea9ba579a3ecdeca0daa313, the password recovery endpoint returns different error messages depending on whether a username exists, so enabling username enumeration. Commit e287fab56089cf8fcea9ba579a3ecdeca0daa313 fixes this issue.

In Plex Media Server (PMS) through 1.42.2.10156, ability to access /myplex/account with a device token is not properly aligned with whether the device is currently associated with an account.

An authentication bypass vulnerability exists in AnythingLLM v1.8.5 in via the /api/workspaces endpoint. The endpoint fails to implement proper authentication checks, allowing unauthenticated remote attackers to enumerate and retrieve detailed information about all configured workspaces. Exposed data includes: workspace identifiers (id, name, slug), AI model configurations (chatProvider, chatModel, agentProvider), system prompts (openAiPrompt), operational parameters (temperature, history length, similarity thresholds), vector search settings, chat modes, and timestamps.

A vulnerability in the mintplex-labs/anything-llm repository, as of commit 5c40419, allows low privilege users to access the sensitive API endpoint "/api/system/custom-models". This access enables them to modify the model's API key and base path, leading to potential API key leakage and denial of service on chats.

In mintplex-labs/anything-llm v1.5.11 desktop version for Windows, the application opens server port 3001 on 0.0.0.0 with no authentication by default. This vulnerability allows an attacker to gain full backend access, enabling them to perform actions such as deleting all data from the workspace.

A vulnerability in AnythingLLM Docker version 1.3.1 allows users with 'Default' permission to access other users' profile pictures by changing the 'id' parameter in the user cookie. This issue is present in versions prior to 1.3.1.

AnythingLLM prior to version 1.3.1 contains a path traversal vulnerability (CWE-22/CWE-29) caused by improper handling of non-ASCII filenames passed through the multer library, where the filename transformation can introduce '../' sequences that are never sanitized. An authenticated attacker holding a manager or admin role can submit a crafted filename to trigger a file write outside the intended upload directory. By writing files to arbitrary server locations, the attacker can escalate to remote code execution on the host running AnythingLLM. All self-hosted AnythingLLM deployments running a version before 1.3.1 are affected, and exploitation requires privileged (manager/admin) application access rather than anonymous access. There is no confirmed in-the-wild exploitation and the issue is not in CISA KEV, but a public proof-of-concept exists via the huntr bounty report and the EPSS score of 21.3% (97th percentile) indicates an elevated likelihood of exploitation within the next 30 days.

Do: Upgrade AnythingLLM to version 1.3.1 or later to obtain the sanitized filename handling. Until upgraded, audit manager and admin accounts for unexpected privileges, review upload activity for filenames containing non-ASCII characters or '../' sequences, and check the host for files written outside the expected upload directories. Internet-facing deployments should be patched with priority given the public PoC and rising EPSS probability.

Deserialization of Untrusted Data vulnerability in PlexTrac (Runbooks modules) which allows Object Injection and arbitrary file writes. This issue affects PlexTrac: from 1.61.3 before 2.8.1.

Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') vulnerability in PlexTrac allows arbitrary file writes.This issue affects PlexTrac: from 1.61.3 before 2.8.1.

mintplex-labs/anything-llm version latest contains a vulnerability where sensitive information, specifically a password, is improperly stored within a JWT (JSON Web Token) used as a bearer token in single user mode. When decoded, the JWT reveals the password in plaintext. This improper storage of sensitive information poses significant security risks, as an attacker who gains access to the JWT can easily decode it and retrieve the password. The issue is fixed in version 1.0.3.

An improper access control vulnerability exists in the mintplex-labs/anything-llm application, specifically within the import endpoint. This vulnerability allows an anonymous attacker, without an account in the application, to import their own database file, leading to the deletion or spoofing of the existing `anythingllm.db` file. By exploiting this vulnerability, attackers can serve malicious data to users or collect information about them. The vulnerability stems from the application's failure to properly restrict access to the data-import functionality, allowing unauthorized database manipulation.

A vulnerability in mintplex-labs/anything-llm allows for a Denial of Service (DoS) condition due to uncontrolled resource consumption. Specifically, the issue arises from the application's failure to limit the size of usernames, enabling attackers to create users with excessively bulky texts in the username field. This exploit results in the user management panel becoming unresponsive, preventing administrators from performing critical user management actions such as editing, suspending, or deleting users. The impact of this vulnerability includes administrative paralysis, compromised security, and operational disruption, as it allows malicious users to perpetuate their presence within the system indefinitely, undermines the system's security posture, and degrades overall system performance.

In mintplex-labs/anything-llm versions up to and including 1.5.3, an issue was discovered where the password hash of a user is returned in the response after login (`POST /api/request-token`) and after account creations (`POST /api/admin/users/new`). This exposure occurs because the entire User object, including the bcrypt password hash, is included in the response sent to the frontend. This practice could potentially lead to sensitive information exposure despite the use of bcrypt, a strong hashing algorithm. It is recommended not to expose any clues about passwords to the frontend.

An uncontrolled resource consumption vulnerability exists in the `upload-link` endpoint of mintplex-labs/anything-llm. This vulnerability allows attackers to cause a denial of service (DOS) by shutting down the server through sending invalid upload requests. Specifically, the server can be made to shut down by sending an empty body with a 'Content-Length: 0' header or by sending a body with arbitrary content, such as 'asdasdasd', with a 'Content-Length: 9' header. The vulnerability is reproducible by users with at least a 'Manager' role, sending a crafted request to any workspace. This issue indicates that a previous fix was not effective in mitigating the vulnerability.

A path traversal vulnerability in mintplex-labs/anything-llm allowed a manager to bypass the `normalizePath()` function, intended to defend against path traversal attacks. This vulnerability enables the manager to read, delete, or overwrite the 'anythingllm.db' database file and other files stored in the 'storage' directory, such as internal communication keys and .env secrets. Exploitation of this vulnerability could lead to application compromise, denial of service (DoS) attacks, and unauthorized admin account takeover. The issue stems from improper validation of user-supplied input in the process of setting a custom logo for the app, which can be manipulated to achieve arbitrary file read, deletion, or overwrite, and to execute a DoS attack by deleting critical files required for the application's operation.

A Cross-Site Scripting (XSS) vulnerability exists in mintplex-labs/anything-llm, affecting both the desktop application version 1.2.0 and the latest version of the web application. The vulnerability arises from the application's feature to fetch and embed content from websites into workspaces, which can be exploited to execute arbitrary JavaScript code. In the desktop application, this flaw can be escalated to Remote Code Execution (RCE) due to insecure application settings, specifically the enabling of 'nodeIntegration' and the disabling of 'contextIsolation' in Electron's webPreferences. The issue has been addressed in version 1.4.2 of the desktop application.

In mintplex-labs/anything-llm, a vulnerability exists in the thread update process that allows users with Default or Manager roles to escalate their privileges to Administrator. The issue arises from improper input validation when handling HTTP POST requests to the endpoint `/workspace/:slug/thread/:threadSlug/update`. Specifically, the application fails to validate or check user input before passing it to the `workspace_thread` Prisma model for execution. This oversight allows attackers to craft a Prisma relation query operation that manipulates the `users` model to change a user's role to admin. Successful exploitation grants attackers the highest level of user privileges, enabling them to see and perform all actions within the system.