A vulnerability in n8n Local Gateway desktop app versions before 2.41.4 and 2.42.x before 2.42.1 allows an attacker to silently re-pair the app with a malicious n8n instance via a crafted deep-link. This can cause the app to connect to an attacker-controlled instance without user confirmation, inheriting previously granted permissions and enabling unauthorized actions through the gateway.
A prototype pollution vulnerability exists in n8n versions before 2.41.4 and in 2.42.x versions before 2.42.1. This flaw allows low-privileged users to manipulate built-in object behavior by saving workflows with specially crafted connection keys such as __proto__. The vulnerability causes server errors during the periodic history compaction task, leading to login failures and write operation disruptions.
n8n versions before 1.123.85, 2.43.5, and 2.44.1 have an authorization bypass vulnerability. This flaw allows authenticated users to access credentials they should not have permission to use by exploiting inline sub-workflow nodes with non-boolean disabled values. Attackers who know a credential ID can execute workflows in inline JSON mode, bypassing permission checks and potentially exfiltrating decrypted credentials to attacker-controlled servers.
n8n versions before 1.123.85, 2.x before 2.43.5, and 2.44.x before 2.44.1 have an OData $filter injection vulnerability in Microsoft Outlook contact lookup and Microsoft SharePoint v1 item update nodes. This flaw allows attackers to inject single quotes into expression-bindable parameters, potentially exposing the entire contact list or redirecting SharePoint writes to unintended list items within the scope of stored credentials.
n8n versions before 1.123.85, 2.x before 2.43.5, and 2.44.x before 2.44.1 contain a stored cross-site scripting (XSS) vulnerability. This occurs because certain webhook response paths, including error responses and responses without headers, do not include the sandbox Content-Security-Policy header. Authenticated users with permissions to create and activate workflows can exploit this to serve malicious HTML that executes scripts in the context of the n8n origin, potentially hijacking user sessions.
A vulnerability in n8n versions before 1.123.85, 2.x before 2.43.5, and 2.44.x before 2.44.1 causes the SeaTable Trigger node's API token to be saved in error execution data. This occurs because failed poll requests store raw request headers, exposing the API token to users who can view workflow executions but do not have credential read access.
n8n versions before 1.123.85, 2.x before 2.43.5, and 2.44.x before 2.44.1 improperly store credential secrets in execution error data for certain nodes. This exposure occurs when operations fail in the Git, MongoDB Atlas Vector Store, and Google Cloud Storage nodes. Users with permission to view workflow executions but without credential read access can extract sensitive secrets such as Git passwords or tokens, MongoDB passwords, and Google access tokens from the error data.
An authorization bypass vulnerability exists in n8n versions before 1.123.85, 2.43.5, and 2.44.1 within the Chat Hub component. This flaw allows users with the global:chatUser role, who normally have limited chat-only permissions, to register restricted tool types and execute sandboxed code, access permitted files, and utilize instance workflows, Data Tables, and agents. The vulnerability has a medium severity rating with a CVSS score of 5.4.
n8n versions prior to 1.123.85, 2.43.5, and 2.44.1 contain an authorization bypass vulnerability. This flaw allows member users to access credentials they should not have permission to use by omitting default node parameters in workflows. Exploitation enables attackers to save and execute workflows that bypass pre-execution credential checks, resulting in decrypted credential values being sent to destinations controlled by the attacker.
A filter injection vulnerability exists in the Quick Base node of n8n versions before 1.123.85, 2.x before 2.43.5, and 2.44.x before 2.44.1. This flaw allows attackers to supply untrusted input, such as webhook data, that is sent as raw query text to the Quick Base API. Exploitation can cause the 'Get Many' operation to return all records or the 'Delete' operation to remove all records accessible by the workflow credentials.
A regular expression denial of service (ReDoS) vulnerability exists in n8n versions before 1.123.85, 2.43.5, and 2.44.1. Authenticated users who can create and run workflows may exploit this by submitting specially crafted wildcard-heavy File(s) Selector patterns. This causes exponential backtracking in the regex engine, blocking the event loop and rendering the UI, API, webhooks, and health checks unresponsive until the service is restarted.
A vulnerability in n8n versions before 2.43.5 and 2.44.x before 2.44.1 allows authenticated members with workflow creation and execution permissions to exploit a client-supplied workflow ID. By using a crafted ID such as __proto__, the application improperly caches workflow owner projects, causing project checks to match all projects. This flaw enables unauthorized listing, reading, modification, and deletion of Data Table rows across all projects within the instance.
A credential guard bypass vulnerability exists in n8n versions before 1.123.85, 2.43.5, and 2.44.1. This flaw allows editors of shared workflows to exploit the Execute Sub-workflow node by loading sub-workflows from URLs or local files, thereby accessing restricted credentials and obtaining their decrypted values.
A vulnerability in chrome-devtools-mcp component of Google Chrome on Linux versions prior to 1.5.0 allows a local attacker to perform arbitrary read and write operations via a malicious file due to insufficient policy enforcement. The issue is rated with medium severity by Chromium security.
Seroval versions 1.6.0 through 1.6.10 contain a vulnerability where typed arrays that have an own constructor property set from untrusted data can lead to code injection during serialization. This occurs because the constructor name is read from a mutable property on the typed array rather than its internal type, allowing arbitrary code to be serialized. The issue is fixed in version 1.6.11. The vulnerability requires unusual conditions where properties from untrusted objects are copied onto typed arrays.
Seroval versions before 1.0.0 and from 1.5.5 up to but not including 1.6.10 improperly serialize RegExp sources containing the '<' character without escaping it. This can allow an attacker to inject script code if the RegExp is built from untrusted input and the serialized output is placed inside an HTML <script> element. The vulnerability is fixed in version 1.6.10 by changing the serialization method to use the RegExp constructor with an escaped string source.
The Seroval library versions prior to 1.6.8 contain a denial-of-service vulnerability in the fromJSON and fromCrossJSON deserialization functions. These functions trust node-supplied array lengths and sequence doneAt indices without validation, allowing crafted inputs to cause infinite loops or prolonged blocking of the event loop. This can be triggered by unauthenticated requests passing untrusted input, resulting in service unavailability. The issue is fixed in version 1.6.8 by enforcing strict validation of array elements and sequence indices, throwing errors on malformed input.
Seroval versions from 1.4.1 up to but not including 1.6.8 contain a denial of service vulnerability in the fromJSON and fromCrossJSON functions. These functions fail to validate that an array node's element list is actually an array, allowing crafted payloads to control memory allocation and processing loops. This can cause the service to become unavailable due to excessive memory use or prolonged thread blocking. The issue bypasses a previous fix and affects any framework using these functions to decode untrusted request bodies before validation. The vulnerability is fixed in version 1.6.8.
A vulnerability in mem0ai's session key construction allows cross-tenant data leakage and corruption. The function _build_session_scope() concatenates user_id, agent_id, and run_id with unescaped '&' and '=' delimiters, enabling an attacker controlling only run_id or agent_id to craft values that collide with another tenant's session key. This leads to unauthorized reading and overwriting of another tenant's conversation history. The flaw affects both Python and TypeScript SDKs and is fixed in versions 2.0.18 and 3.1.6.
A vulnerability in ONNX ConvTranspose shape inference causes an out-of-bounds read when the weight tensor has fewer than three dimensions and no kernel_shape attribute is provided. This leads to a process crash (denial of service) during shape inference on a crafted model. The issue affects ONNX versions 1.2.1 through 1.21.0 and is fixed in version 1.22.0. No confidentiality or integrity impact is reported.