Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, any authenticated user can permanently delete files owned by other users via DELETE /api/v1/files/{id} when the target file is referenced in any shared chat. The has_access_to_file() authorization gate unconditionally grants access through its shared-chat branch. It checks neither the requesting user's identity nor the type of operation being performed. File UUIDs (which would otherwise be impractical to guess) are disclosed to any user with read access to a knowledge base via GET /api/v1/knowledge/{id}/files. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, any authenticated user with low privileges can enumerate active background tasks across the system and stop tasks belonging to other users via the GET /api/tasks and POST /api/tasks/stop/{task_id} methods. This allows a casual user to disrupt system-wide chat usage by continuously canceling other users' active tasks. This is a real authorization vulnerability affecting integrity and usability in multi-user deployments. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, a user just needs to use the API endpoint: /api/chat/completions with their own API key (generated in OWUI) and the Chat ID of another user to continue the conversation of the other user. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, validate_url() in backend/open_webui/retrieval/web/utils.py calls validators.ipv6(ip, private=True), but the validators library does NOT implement the private keyword for IPv6 — the call raises a ValidationError (which is falsy in a boolean context), so every IPv6 address passes the filter. In addition, IPv4-mapped IPv6 (::ffff:10.0.0.1) bypasses the IPv4 check entirely, and several reserved IPv4 ranges (0.0.0.0/8, 100.64.0.0/10, 192.0.0.0/24, etc.) are not blocked. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, the /responses endpoint in the OpenAI router accepts any authenticated user and forwards requests directly to upstream LLM providers without enforcing per-model access control. While the primary chat completion endpoint (generate_chat_completion) checks model ownership, group membership, and AccessGrants before allowing a request, the /responses proxy only validates that the user has a valid session via get_verified_user. This allows any authenticated user to interact with any model configured on the instance by sending a POST request to /api/openai/responses with an arbitrary model ID. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, Open WebUI supports model composition via base_model_id: a user-defined model (e.g., "Cheap Assistant") can reference an existing base model (e.g., "gpt-4-turbo-restricted") that provides the actual inference capability. When a user queries the composed model, the access control pipeline verifies the user has access to the composed model but never re-verifies access to the chained base model. Additionally, the model creation and import endpoints accept arbitrary base_model_id values without checking that the caller has access to that base model. Combined, this allows any user with the default model creation permission to create a model that chains to a restricted base model — and then invoke it, causing the server to dispatch the request to the restricted base model using the admin-configured API key. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, the POST /api/v1/retrieval/process/web endpoint accepts a user-supplied collection_name and an overwrite query parameter (default: True). It performs no authorization check on whether the calling user owns or has write access to the target collection. When overwrite=True, save_docs_to_vector_db calls VECTOR_DB_CLIENT.delete_collection() on the target collection before writing new content. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, administrative role changes and user deletions do not iterate SESSION_POOL to disconnect affected sessions. As a result, a user whose admin role has been revoked retains admin privileges within their existing Socket.IO session for as long as they keep the connection alive (via automatic heartbeats). The gap is exclusive to the Socket.IO session cache. This vulnerability is fixed in 0.9.0.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, the tool_servers and terminal_servers keys in utils/tools.py do use a prefix. When two or more Open WebUI instances share a Redis database (a supported and documented deployment pattern, e.g., for multi-region deployments, blue-green setups, or cluster topologies), the unprefixed keys collide. An admin on Instance A writing to tool_servers overwrites the value read by Instance B — causing Instance B's users to receive Instance A's tool server configuration. This vulnerability is fixed in 0.9.0.
Missing bounds validation in the MQTT v5.0 property parser in coreMQTT before 5.0.1 allows an MQTT broker to cause a denial of service by sending a crafted packet.
To remediate this issue, users should upgrade to v5.0.1.
Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.3, the checkout endpoint accepts a user-controlled cart_id and uses it to enter the payment flow without verifying cart ownership. A logged-in attacker can therefore reuse another user's cart data in their own checkout session. This vulnerability is fixed in 1.0.8.3.
Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.3, the backend admin/auth-token endpoint allows an authenticated administrator to load another administrator's REST API token list by supplying that user's admin_id. This can disclose sensitive API tokens belonging to other administrators. This vulnerability is fixed in 1.0.8.3.
phpMyFAQ before 4.1.2 contains a stored cross-site scripting vulnerability in Utils::parseUrl() that allows authenticated users to inject JavaScript via malformed URLs in comments. Attackers can craft URLs with unescaped quotes to inject event handlers, stealing admin session cookies and achieving full application takeover when visitors view affected FAQ pages.
phpMyFAQ before 4.1.2 contains an information disclosure vulnerability in the getIdFromSolutionId() method that lacks permission filtering, allowing unauthenticated attackers to enumerate restricted FAQ entries and read their titles via the /solution_id_{id}.html endpoint. Attackers can sequentially iterate solution IDs to discover all FAQs including those restricted to specific users or groups, leaking sensitive metadata through redirect Location headers and page canonical links.
phpMyFAQ before 4.1.2 contains a sql injection vulnerability in CurrentUser::setTokenData that allows authenticated attackers to execute arbitrary SQL by injecting malicious OAuth token claims. Attackers with Azure AD accounts containing SQL metacharacters in display names or JWT claims can break out of string literals and execute arbitrary database queries.
Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.2, Vvveb CMS does not validate the sign of the quantity parameter on the cart-add endpoint. Submitting a negative integer is accepted by the server and treated as a normal positive line-item, but with the sign carried through into every downstream computation: line total, sub-total, taxes, and grand total all become negative numbers. The customer-facing cart UI then displays a negative grand total to the user, the checkout flow accepts the negative cart, and the resulting order is persisted in the merchant's database with a negative total column. From the merchant's order management dashboard, this surfaces as a real order with a negative total — an "the merchant owes the customer money" record that no legitimate workflow ever creates. This vulnerability is fixed in 1.0.8.2.
PHP Timeclock 1.04 contains time-based and boolean-based blind SQL injection vulnerabilities in the login_userid parameter of login.php that allows unauthenticated attackers to extract database contents. Attackers can submit crafted POST requests with SQL payloads using SLEEP functions or RLIKE conditional statements to dump sensitive database information including employee names and credentials.
Schlix CMS 2.2.6-6 contains a remote code execution vulnerability that allows authenticated attackers to execute arbitrary PHP code by uploading malicious extension packages through the block manager. Attackers can upload a crafted ZIP file containing PHP code in the packageinfo.inc file and trigger execution by accessing the About tab of the installed extension.
Anote 1.0 contains a persistent cross-site scripting vulnerability that allows attackers to execute arbitrary code by injecting malicious payloads into markdown files stored within the application. Attackers can craft malicious markdown files with embedded JavaScript that executes system commands when opened, enabling remote code execution on the victim's computer.
WordPress Plugin WPGraphQL 1.3.5 contains a denial of service vulnerability that allows unauthenticated attackers to exhaust server resources by sending batched GraphQL queries with duplicated fields. Attackers can send POST requests to the GraphQL endpoint with amplified field duplication payloads to trigger server out-of-memory conditions and MySQL connection errors.
Trog::TOTP versions before 1.006 for Perl generate secrets using rand.
Secrets were generated using Perl's built-in rand function, which is predictable and unsuitable for security usage.
radare2 6.1.5 contains a use-after-free vulnerability in the gdbr_threads_list() function that allows remote attackers to trigger memory corruption by sending a valid qfThreadInfo response followed by a malformed qsThreadInfo response. Attackers can exploit this vulnerability through GDB remote debugging to cause a denial of service or potentially achieve code execution by manipulating thread list processing.
Microsoft APM is an open-source, community-driven dependency manager for AI agents. From 0.5.4 to 0.12.4, two primitive integrators in apm-cli enumerate package files with bare Path.glob() / Path.rglob() calls and read each match with Path.read_text(), transparently following symbolic links. A symlink committed inside a remote APM dependency under .apm/prompts/<x>.prompt.md or .apm/agents/<x>.agent.md is preserved verbatim into apm_modules/ on clone and then dereferenced during integration, with the resolved content written as a regular file into the project's deploy directories. The package content_hash, the pre-deploy SecurityGate scan, and apm audit do not flag this. The deploy roots are not added to the auto-generated .gitignore, so the resulting files are staged by git add by default. This vulnerability is fixed in 0.13.0.
Tabby (formerly Terminus) is a highly configurable terminal emulator. Prior to 1.0.233, since Tabby does not escape control characters from file paths when dragging and dropping a file into it, code execution can be achieved. This vulnerability is fixed in 1.0.233.
Tabby (formerly Terminus) is a highly configurable terminal emulator. Prior to 1.0.232, Tabby's terminal linkifier passes any detected URI directly to the operating system's protocol handler without validating the protocol scheme. This allows a malicious SSH or Telnet server to send crafted terminal output containing dangerous protocol URIs which Tabby renders as clickable links, triggering arbitrary OS protocol handlers on the victim's machine. This vulnerability is fixed in 1.0.232.
Tabby (formerly Terminus) is a highly configurable terminal emulator. Prior to 1.0.233, Tabby before 1.0.233 automatically confirms ZMODEM protocol detection on all terminal session output without user interaction, enabling shell command execution when a user displays attacker-controlled content. The ZModemMiddleware in tabby-terminal consumes all session output through a Zmodem.Sentry, and when a ZMODEM ZRQINIT header is detected, unconditionally calls detection.confirm() and writes a fixed ZRINIT response ( **\x18B0100000023be50\r\n\x11) back into the active PTY as input. When the process that triggered the detection (e.g., cat) exits, the injected bytes are consumed by the user's shell as a command line. Under fish (default configuration), the ** prefix triggers recursive glob expansion against the current directory, allowing an attacker-placed executable at a matching nested path (e.g., d/xB0100000023be50) to be executed by relative pathname without relying on PATH. Under bash and zsh, a secondary xterm.js terminal color-query feedback (OSC 10) can be combined in the same file to inject a slash-containing command word that similarly bypasses PATH resolution. An attacker can exploit this by providing a crafted file (e.g., in a cloned Git repository) that a user displays with cat, achieving code execution with no interaction beyond viewing the file. This vulnerability is fixed in 1.0.233.
Tabby (formerly Terminus) is a highly configurable terminal emulator. Prior to 1.0.233, Tabby registers itself as the handler for the tabby:// URL scheme on all platforms. The URL scheme handler supports a run command that directly executes OS commands with no user confirmation, sanitization, or sandboxing. An attacker can craft a malicious link (tabby://run?command=...) and deliver it via a website, email, chat message, or any other medium. When a victim clicks the link, the OS launches Tabby which immediately spawns the specified command as a child process with the user's full privileges. This is a zero-click-after-link-visit RCE vulnerability. This vulnerability is fixed in 1.0.233.
The bitcoinj library is a Java implementation of the Bitcoin protocol. Prior to 0.17.1, ScriptExecution.correctlySpends() contains two fast-path verification bugs for standard P2PKH and native P2WPKH spends in core/src/main/java/org/bitcoinj/script/ScriptExecution.java. In both branches, bitcoinj verifies an attacker-controlled signature/public-key pair but fails to verify that the public key is the one committed to by the output being spent. As a result, any attacker keypair can satisfy bitcoinj's local verification for arbitrary P2PKH and P2WPKH outputs. This vulnerability is fixed in 0.17.1.
Microsoft APM is an open-source, community-driven dependency manager for AI agents. Prior to 0.8.12, Microsoft APM normalizes marketplace plugins by copying plugin components referenced in plugin.json into .apm/. The manifest fields agents, skills, commands, and hooks are attacker-controlled, but the implementation does not enforce that those paths remain inside the plugin directory. A malicious plugin can therefore use absolute paths or ../ traversal paths to copy arbitrary readable host files or directories from the installer's machine during apm install. This vulnerability is fixed in 0.8.12.
Turborepo is a high-performance build system for JavaScript and TypeScript codebases. Prior to 2.9.14000, the Turborepo LSP VS Code extension could execute shell commands derived from workspace-controlled values. The extension used string-based command execution for Turborepo daemon commands and task runs. A malicious workspace could provide crafted values through workspace settings or task names in the repository's source code that were interpolated into shell commands. When the extension activated or when a user ran a task through the extension, those values could be interpreted by the user's shell, allowing arbitrary command execution with the privileges of the local VS Code process. This vulnerability is fixed in 2.9.14000.
Code injection in SQL code generation in Apache Flink 1.15.0 through 1.20.x and 2.0.0 through 2.x allows authenticated users with query submission privileges to execute arbitrary code on TaskManagers via maliciously crafted SQL queries. The vulnerability affects JSON functions (1.15.0+) and LIKE expressions with ESCAPE clauses (1.17.0+). User-controlled strings are interpolated into generated Java code without proper escaping, allowing attackers to break out of string literals and inject arbitrary expressions.
Users are recommended to upgrade to either version 1.20.4, 2.0.2, 2.1.2 or 2.2.1, which fixes this issue.
Oinone Pamirs 7.0.0 contains a command injection vulnerability in CommandHelper.executeCommands. The method starts a shell process and writes attacker-controlled command strings directly to the process standard input without sanitization. In affected deployments, this can result in arbitrary operating system command execution.
An issue in Nodemailer smtp_server before v.3.18.3 allows a remote attacker to cause a denial of service via the SMTPStream._write, lib/smtp-stream.js components
A buffer underflow vulnerability has been identified in the ogg123 utility from the vorbis-tools 1.4.3 package in function remotethread in remote.c. This vulnerability occurs in the remote control functionality when processing malformed input, leading to a stack buffer underflow that can cause application crashes and potentially allow code execution.
In the Linux kernel, the following vulnerability has been resolved:
ptrace: slightly saner 'get_dumpable()' logic
The 'dumpability' of a task is fundamentally about the memory image of
the task - the concept comes from whether it can core dump or not - and
makes no sense when you don't have an associated mm.
And almost all users do in fact use it only for the case where the task
has a mm pointer.
But we have one odd special case: ptrace_may_access() uses 'dumpable' to
check various other things entirely independently of the MM (typically
explicitly using flags like PTRACE_MODE_READ_FSCREDS). Including for
threads that no longer have a VM (and maybe never did, like most kernel
threads).
It's not what this flag was designed for, but it is what it is.
The ptrace code does check that the uid/gid matches, so you do have to
be uid-0 to see kernel thread details, but this means that the
traditional "drop capabilities" model doesn't make any difference for
this all.
Make it all make a *bit* more sense by saying that if you don't have a
MM pointer, we'll use a cached "last dumpability" flag if the thread
ever had a MM (it will be zero for kernel threads since it is never
set), and require a proper CAP_SYS_PTRACE capability to override.
PDF Export Module used in DHTMLX's products Gantt and Scheduler is vulnerable to Path Traversal due to lack of HTML sanitization. An unauthenticated user could craft the html payload which could include
local files from the server and display them in the generated PDF.
This issue was fixed in PDF Export Module version 0.7.6.
The Quick Playground plugin for WordPress is vulnerable to Path Traversal in versions up to and including 1.3.3. This is due to insufficient path validation in the qckply_zip_theme() function, which appends a user-controlled 'stylesheet' parameter directly to the theme root directory path without sanitizing directory traversal sequences. This makes it possible for unauthenticated attackers to trigger the creation of a ZIP archive containing arbitrary files from the server's filesystem — including wp-config.
The Frontend Admin by DynamiApps plugin for WordPress is vulnerable to Privilege Escalation in versions up to and including 3.28.36. This is due to insufficient authorization checks in the role field update mechanism combined with overly permissive capabilities for the admin_form post type. The admin_form custom post type uses 'capability_type' => 'page', which grants editors the ability to create and edit forms. When an editor creates an edit_user form, they can manipulate the form configuration to include 'administrator' in the role_options array by directly submitting POST data to wp-admin/post.php, bypassing the UI restrictions in feadmin_get_user_roles(). When the form is subsequently submitted, the pre_update_value() function in class-role.php only validates that the submitted role exists in the form's role_options array (lines 107-110), but fails to verify that the current user has permission to assign that specific role. This makes it possible for unauthenticated attackers to first register as editors (via a public new_user form), then create an edit_user form with administrator in the allowed roles, and finally use that form to escalate their own privileges to administrator.
The FOX – Currency Switcher Professional for WooCommerce plugin for WordPress is vulnerable to unauthorized data loss due to a missing capability check on the 'admin_head' function in all versions up to, and including, 1.4.5. This makes it possible for authenticated attackers, with Contributor-level access and above, to delete the entire multi-currency configuration by visiting any wp-admin page with the `woocs_reset` parameter appended. Additionally, because no nonce is verified, this is also exploitable via Cross-Site Request Forgery against any administrator. The vulnerability may also be exploited by Subscriber-level users if the site is configured to allow Subscriber access to 'wp-admin' pages.
VMware Fusion contains a TOCTOU (Time-of-check Time-of-use) vulnerability that occurs during an operation performed by a SETUID binary. A malicious actor with local non-administrative user privileges may exploit this vulnerability to escalate privileges to root on the system where Fusion is installed.
In the Linux kernel, the following vulnerability has been resolved:
ksmbd: validate inherited ACE SID length
smb_inherit_dacl() walks the parent directory DACL loaded from the
security descriptor xattr. It verifies that each ACE contains the fixed
SID header before using it, but does not verify that the variable-length
SID described by sid.num_subauth is fully contained in the ACE.
A malformed inheritable ACE can advertise more subauthorities than are
present in the ACE. compare_sids() may then read past the ACE.
smb_set_ace() also clamps the copied destination SID, but used the
unchecked source SID count to compute the inherited ACE size. That could
advance the temporary inherited ACE buffer pointer and nt_size accounting
past the allocated buffer.
Fix this by validating the parent ACE SID count and SID length before
using the SID during inheritance. Compute the inherited ACE size from the
copied SID so the size matches the bounded destination SID. Reject the
inherited DACL if size accumulation would overflow smb_acl.size or the
security descriptor allocation size.
Cross-site request forgery vulnerability exists in Musetheque V4 Information Disclosure for IPKNOWLEDGE V4L1 rev2203.0 and earlier. If a user views a malicious page while logged-in to the affected product, unexpected operations may be done.
A DLL hijacking vulnerability in the AMD Cleanup Utility could allow an attacker to achieve privilege escalation potentially resulting in arbitrary code execution.
A vulnerability in mlflow/mlflow versions 3.9.0 and earlier allows unauthenticated access to certain FastAPI routes when the server is started with authentication enabled (`--app-name basic-auth`) and served via uvicorn (ASGI). The FastAPI permission middleware only enforces authentication on `/gateway/` routes, leaving other routes such as the Job API (`/ajax-api/3.0/jobs/*`) and the OpenTelemetry trace ingestion API (`/v1/traces`) unprotected. This allows unauthenticated remote attackers to submit jobs, read job results, cancel running jobs, and inject arbitrary trace data into experiments. The issue arises from an architectural mismatch between Flask and FastAPI authentication mechanisms, where the `_find_fastapi_validator()` function fails to handle non-`/gateway/` paths, resulting in a complete authentication bypass. This vulnerability is fixed in version 3.10.0.
ZITADEL is an open source identity management platform. From 2.71.11 to before 3.4.10 and 4.15.0, a vulnerability was discovered in Zitadel's LDAP identity provider implementation, which fails to properly escape user-provided usernames before incorporating them into LDAP search filters. This allows unauthenticated attackers to perform LDAP Filter Injection during the login process. While this vulnerability does not allow for a full authentication bypass, an attacker can use LDAP metacharacters (such as *, (, )) to perform blind LDAP injection. By observing the different failure (or success) responses, an attacker can systematically enumerate valid usernames and extract sensitive attribute data from the connected LDAP directory. This vulnerability is fixed in 3.4.10 and 4.15.0.
python-utcp is the python implementation of UTCP. Prior to 1.1.3, _prepare_environment() in cli_communication_protocol.py passes a full copy of os.environ to every CLI subprocess. When combined with CVE-2026-45369, an attacker can exfiltrate all process-level secrets in a single tool call. This vulnerability is fixed in 1.1.3.
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.
libyang is a YANG data modeling language library. Prior to SO 5.2.15, lyb_read_string() in src/parser_lyb.c contains an integer overflow that results in a heap buffer overflow when parsing a maliciously crafted LYB binary blob. An attacker who can supply LYB data to any libyang consumer (NETCONF server, sysrepo, etc.) can trigger a crash or potential heap corruption. This vulnerability is fixed in SO 5.2.15.
Crabbox prior to v0.12.0 contains a privilege escalation vulnerability that allows users with shared visibility-only access to obtain Code, WebVNC, and Egress agent tickets by sending POST requests to ticket endpoints. Attackers can exploit insufficient access control checks on the /v1/leases/:id/code/ticket, /v1/leases/:id/webvnc/ticket, and /v1/leases/:id/egress/ticket endpoints to obtain bridge-agent tickets and impersonate trusted lease-side bridges despite having only visibility permissions.