NocoDB versions prior to 2026.10.1 contain a stored CSS injection vulnerability that allows authenticated users with Commenter or Editor roles to inject inline styles via single quotes in mention text. This injection can manipulate the user interface for other viewers by breaking out of the single-quoted data-id attribute in comments or rich text cells.
vLLM from 0.6.0 before 0.33.0 lacks a request-wide media budget, allowing remote inference clients to exhaust frontend resources by submitting many remote image URLs in one request. Attackers can send a compact request to /v1/messages or /v1/responses that forces downloading and decoding of every image before prompt-length admission, exhausting memory, CPU, and connections.
vLLM before 0.33.0 contains a path traversal vulnerability that allows malicious model publishers to redirect the served model and tokenizer by setting speculators_config verifier.name_or_path in config.json. Attackers can point this field at another local checkpoint directory to leak its tokenizer vocabulary via unauthenticated /detokenize requests and gain query access to private models.
vLLM from 0.28.0 before 0.33.0 contains an improper input validation vulnerability that allows unauthenticated attackers to crash the shared EngineCore by sending out-of-range prompt token ids when served with --skip-tokenizer-init. Attackers can submit a large positive token id to /v1/completions, /v1/embeddings, or /pooling, bypassing the tokenizer-gated vocab check and causing a fatal embedding index error that denies service to all tenants.
CVE-2026-130683 is a critical path traversal vulnerability in siyuan-note versions before 3.8.10. The application uses an incomplete deny-list for sensitive paths, omitting important shell initialization files and directories. Authenticated administrators can exploit this flaw by calling the /api/file/copyFile endpoint with an absolute path outside the workspace, allowing them to write malicious payloads that execute with user privileges upon the next shell or login. This vulnerability is fixed in version 3.8.10.
SiYuan before 3.8.10 persists a table block's caption attribute into IAL properties without HTML sanitization, causing stored cross-site scripting upon import or sync. Attackers can inject an entity-encoded caption with an onerror handler that executes in the Electron client's nodeIntegration context to achieve remote code execution.
SiYuan before 3.8.10 contains a stored cross-site scripting vulnerability, an incomplete fix for CVE-2026-44670, because renderAVAttribute() inserts an Attribute View name unescaped into the database row attribute panel header. Attackers who can persist the name, such as via the setAttrViewName action, can inject an onerror handler that runs OS commands through Node.js when a desktop user opens the panel.
SiYuan from 3.3.0 before 3.8.10 contains a stored cross-site scripting vulnerability that allows attackers to inject HTML via unescaped checkbox column names in Gallery and Kanban AttributeView renderers. Attackers can share a crafted .sy.zip document whose rendering in the Electron desktop client with nodeIntegration enabled executes Node.js and host OS commands.
SiYuan before 3.8.10 contains a stored cross-site scripting vulnerability that allows attackers to execute script by storing malicious Attribute View column names rendered unescaped in col.ts. Attackers can inject HTML via addAttributeViewKey, .sy.zip import, or sync, and when a victim adds or duplicates the column, insertAdjacentHTML runs JavaScript with Node.js privileges for desktop code execution.
SiYuan note-taking software versions before 3.8.10 contain a stored cross-site scripting (XSS) vulnerability in the genAVValueHTML() function. This flaw allows attackers who can store Attribute View asset values to inject malicious scripts via an unescaped image src attribute. The injected script can include a double quote and an onerror handler that executes operating system commands through Node.js when a user opens the affected note or attribute panel in the desktop application. This vulnerability is critical and has been fixed in version 3.8.10.
Craft CMS from 5.0.0 before 5.11.4 contains an improper session management vulnerability that fails to invalidate existing sessions when 2FA is enabled and allows disabling 2FA without re-authentication. Attackers holding a hijacked or stale session can retain full account access and silently remove 2FA without the password, authenticator code, or recovery code.
Craft CMS from 5.0.0 before 5.11.4 contains a stored cross-site scripting vulnerability that allows low-privilege control panel users to inject JavaScript by poisoning cached license data. Attackers can store a malicious licensed domain via the app/process-api-response-headers endpoint, which executes unescaped in administrators' dashboard license-mismatch warnings.
Craft CMS versions from 5.0.0 up to but not including 5.11.4 contain a critical remote code execution vulnerability in the ImageTransforms::normalizeTransform() function. This flaw allows attackers to instantiate arbitrary PHP classes by supplying a crafted class:null parameter with a __class key. Exploitation can occur via Craft Commerce's payments/pay redirect template when calling customer.photo.getUrl() with a malicious transform, enabling execution of classes such as yii\rbac\PhpManager and arbitrary PHP code execution. The vulnerability is fixed in version 5.11.4.
Craft CMS from 5.0.0 before 5.11.4 contains an incomplete blocklist fix that allows attackers who can author entry type title formats to write arbitrary files using ZipArchive or XMLWriter classes. Attackers can abuse the non-sandboxed renderObjectTemplate() renderer to write attacker-controlled content into the web docroot, achieving remote code execution.
Craft CMS from 5.0.0 before 5.11.5 contains an sql injection vulnerability that allows authenticated low-privilege users to inject raw SQL by passing nested array parameters into Element::findOne() criteria. Attackers can set the query's where property through endpoints like users/delete-address and users/save-address to perform blind boolean or time-based injection and read database contents.
Craft CMS from 5.0.0 before 5.11.5 contains a missing authorization vulnerability in ChartsController::actionGetNewUsersData() that lacks a viewUsers permission check. Authenticated control panel users can POST to the charts/get-new-users-data action to read aggregate user-registration time series and totals for the site or any user group.
Parse Server versions before 8.6.119 and 9.x before 9.10.8-alpha.3 have a missing authorization vulnerability. This flaw allows LiveQuery clients without the master key to subscribe to installation and internal classes. Exploiting this, attackers can access installation IDs and device tokens of other devices and modify installations to redirect push notifications.
ImageMagick before 7.1.2-35 contains a heap buffer over-read vulnerability that allows attackers to read beyond allocated memory when BGRO image data is imported. Attackers can supply crafted BGRO data for processing to crash the application or leak adjacent heap memory contents.
ImageMagick before 7.1.2-35 contains a null pointer dereference vulnerability that allows attackers to crash the process by using the -subimage option on a single-image list. Attackers can supply input producing an image list with only one image while -subimage is applied, causing a denial of service.
ImageMagick before 7.1.2-35 contains a double free vulnerability that allows attackers to trigger an invalid memory free by using the DPC compare metric. Attackers can supply images for comparison with the DPC metric to corrupt memory and crash the application, causing denial of service.