CVE Feed

Mooncake transfer engine before 0.3.13 contains an untrusted pointer dereference in ServerSession::readHeader that allows unauthenticated attackers to read and write arbitrary process memory via the TCP transport data port. Attackers can send a crafted SessionHeader with arbitrary addr and size values using READ or WRITE opcodes to disclose KV cache contents, prompts and secrets or corrupt memory toward code execution.

Mooncake transfer engine through 0.3.13.post1 contains a memory exhaustion vulnerability in TransferMetadata::receivePeerNotify that allows unauthenticated attackers to grow process memory without limit. Attackers can repeatedly send notify frames up to 1 MB to the handshake RPC port, filling the uncapped notifys vector until the out-of-memory killer terminates the engine.

Mooncake transfer engine through 0.3.13.post1 contains a denial of service vulnerability that allows unauthenticated remote attackers to block the handshake daemon by never reading replies. Attackers can send a Metadata request to the handshake RPC port and stall SocketHandShakePlugin's single listener thread in writeFully(), breaking all subsequent handshakes, metadata fetches, notify and probe requests.

Office-PowerPoint-MCP-Server through 2.0.7 contains a path traversal vulnerability that allows MCP callers to write and read files outside the working directory by supplying absolute paths or ../ sequences. Attackers can steer an AI agent via prompt injection to abuse save_presentation, open_presentation, or manage_image output_path to overwrite any server-writable file or load external files.

This vulnerability enables unauthenticated remote code execution (RCE) on a victim's machine by exploiting a combination of cryptographic weaknesses and memory management issues in the SConnect native host component. The attack leverages an unrestricted messaging interface between an attacker-controlled web page and the native host, allowing malicious input to bypass security checks.

- Use of Hard-coded Credentials vulnerability in Johnson Controls EasyIO FG allows - Pasword Spraying. This issue affects EasyIO FG: before 2.0b52.

- Cleartext Transmission of Sensitive Information vulnerability in Johnson Controls EasyIO NEO allows - Man In the Middle Attack. This issue affects EasyIO NEO: before 3.3b25.

- Exposure of Sensitive Information vulnerability in Johnson Controls Easy IO Neo allows Collect Data from Common Resource Locations. This issue affects Easy IO Neo: before 3.3b63.

PictShare before version 3.7.1 contains a weak randomness vulnerability where the getRandomString() function uses the non-cryptographic rand() PRNG to generate the delete_code authorization token in src/inc/core.php. Attackers can predict or infer the PRNG state to guess valid delete_code values and perform unauthorized deletion of hosted files without needing to read the code from the info endpoint.

- On-Chip Debug Interface vulnerability in Johnson Controls Neo Series MVP2 allows Collect Data from Common Resource Locations. This issue affects Neo Series MVP2: before 3.3b63.

- On-Chip Debug Interface vulnerability in Johnson Controls EasyIO FS32 allows Collect Data from Common Resource Locations. This issue affects EasyIO FS32: before 3.3b63.

: Use of Hard-coded Cryptographic Key vulnerability in Johnson Controls EasyIO FS32 allows : Retrieve Embedded Sensitive Data. This issue affects EasyIO FS32: before 3.0b63.

: Insecure Default Initialization of Resource vulnerability in Johnson Controls EasyIO FS32 allows : Authentication Abuse. This issue affects EasyIO FS32: before 3.0b63.

Improper neutralization of input during web page generation ('cross-site scripting') vulnerability in CWE-79 - Cross-site Scripting CAPEC-63 allows Cross-Site Scripting (XSS). This issue affects CAPEC-63: before 3.0b63.

- External Control of File Name or Path vulnerability in Johnson Controls EasyIO FS32 allows - traversal attack. This issue affects EasyIO FS32: before 3.0b63.

PictShare before 3.7.1 contains an information disclosure vulnerability that allows unauthenticated attackers to obtain the secret delete_code and uploader metadata by calling the API::info() endpoint which returns the complete raw metadata object without a field whitelist. Attackers can use the publicly visible file hash to retrieve the delete_code via the info API and then invoke the delete API to permanently delete arbitrary files, while also exposing uploader IP, User Agent, remote port, and SHA-1 hash, resulting in loss of content integrity, availability, and uploader privacy.

- OS Command Injection vulnerability in Johnson Controls EasyIO FS32 allows OS Command Injection. This issue affects EasyIO FS32: before 3.0b63.

A fail-open error handling issue within the data masking utility of Powertools for AWS Lambda (Python) might allow actors to read sensitive field values that the application intended to mask.  To remediate this issue, users should upgrade to version 3.35.0.

- OS Command Injection vulnerability in Johnson Controls EasyIO FS32 allows - Command Injection. This issue affects EasyIO FS32: before 3.0b63.

: Use of Hard-coded Credentials vulnerability in Johnson Controls EasyIO FS32 allows : Exploitation of Default or Hard-coded Credentials. This issue affects EasyIO FS32: before 3.0b63.

Kasa EC70 v4 and EC71 v4 do not logically disable the production debug interface at the firmware or chip level and do not lock the bootloader.  Although the debug traces are physically severed during manufacturing, an attacker with physical access can restore the connection, interrupt the boot process, and manipulate boot parameters to enter a non-standard initialization path that exposes an unauthenticated root shell during startup. Successful exploitation may allow an attacker with physical access to obtain root-level command access during device startup, resulting in loss of confidentiality, integrity, and availability for the affected device. Exploitation requires device disassembly, restoration of the severed debug connection, and manipulation of the boot process.

Uncontrolled recursion in the Ion reader in Amazon Ion Python before 0.15.0 might allow a remote unauthenticated actor to crash the application using the library, resulting in a denial of service, via a crafted, deeply nested Ion value. To remediate this issue, users should upgrade to version 0.15.0 or later.

figlet.js is a FIG driver written in JavaScript that aims to implement the FIGfont specification. Prior to 1.11.3, text() and textSync() can enter an unbounded loop when whitespaceBreak is enabled and width is smaller than the rendered width of a single FIGlet character. Under these conditions, breakWord() cannot find a valid break point and returns without consuming a character, so generateFigTextLines() repeatedly processes the same input while consuming CPU and growing memory. The non-default option and attacker-controlled width must both reach an affected call. This issue is fixed in version 1.11.3.

stream-json is a micro-library of stream components for processing JSON and JSONC with a minimal memory footprint. Prior to 3.6.0, Assembler materializes object properties with plain assignment, so an input key named __proto__ invokes the inherited setter and causes parsed object prototype replacement instead of creating an own data property. Applications that make authorization or feature decisions from inherited values can therefore consume attacker-controlled properties, and a null prototype can disrupt code that expects Object.prototype methods. The researcher treats parsing untrusted JSON as part of the project contract, while the maintainer states that documented inputs are locally owned dumps, exports, or logs and characterizes the attack vector as local. The global Object.prototype is not polluted. This issue is fixed in version 3.6.0.

stream-json is a micro-library of stream components for processing JSON and JSONC with a minimal memory footprint. Prior to 3.6.0, the JSONC parser at stream-json/jsonc/parser.js and verifier at stream-json/jsonc/verifier.js restart comment-terminator scanning from the opening slash whenever a block or line comment spans an input chunk, while retaining the accumulated comment buffer. Delivering a large valid comment across many small chunks therefore causes quadratic CPU work and can stall the Node.js event loop. The maintainer characterizes the attack vector as local because the documented JSONC input is locally owned or user-controlled configuration, rather than input intended for the open internet. This JSONC-only scope does not include the plain JSON parser, which advances through and discards consumed string and number data. This issue is fixed in version 3.6.0.

Filament is a collection of full-stack components for accelerated Laravel development. From 4.0.0 until 4.13.3 and 5.8.3, app-based multi-factor authentication management actions do not consistently require confirmation of the current password. An attacker with access to an authenticated user session can set up app-based MFA and obtain recovery codes, or disable app-based MFA and regenerate recovery codes by supplying an existing app code or recovery code, without knowing the account password. Email-based MFA is not affected, and the issue does not independently permit an unauthenticated sign-in, but changing the app-MFA configuration may lock the legitimate user out. This issue is fixed in versions 4.13.3 and 5.8.3.

Out-of-bounds Write (CWE-787) in the PEA archive extraction routine (pea.pas, unpea_procedure) of the first-party pea component in PeaZip 11.2.0 and earlier allows an attacker who convinces a victim to open or extract a crafted .pea archive to execute arbitrary code as the user running PeaZip. While decompressing a PCOMPRESS1 stream, the 32-bit compressed-block-size field of the first block (compsize) is read directly from the archive and used without validation as the length of a blockread into the fixed-size global buffers wbuf1/wbuf2 (1,114,112 bytes) and as the bound of the subsequent copy loop. The existing check "compsize > WBUFSIZE" is applied only to the size of each following block, so the first block escapes it; the same unvalidated value is also used to index wbuf1[compsize], an out-of-bounds read at an attacker-chosen offset. The copy loop additionally copies the requested length instead of the number of bytes actually read, and terminates on equality rather than on an upper bound. Because the project is built without range checking and no archive password, integrity tag or non-default configuration is required, the overflow overwrites adjacent global data; code execution was demonstrated by two independent researchers against the official Linux x86-64 and Windows x64 builds, and the denial-of-service and memory-corruption primitive is cross-platform (Windows, macOS, Linux, BSD).

Cleartext transmission without a cryptographic integrity check in operator control unit to robot UDP traffic in Teledyne FLIR Aware2 versions through 6.9.0.2 (PackBot) and 1.7.9 (FirstLook) allows adjacent unauthenticated attackers to intercept, hijack, or modify control traffic against Teledyne FLIR PackBot and FirstLook robots running this software via sniffing or hijacking network traffic.

Hardcoded passwords in the access control in Teledyne FLIR Aware2 versions through 6.9.0.2 (PackBot) and 1.7.9 (FirstLook) allows remote unauthenticated attackers to access and reconfigure Teledyne FLIR PackBot and FirstLook robots running this software via reading the passwords from the firmware or documentation.

Unencrypted traffic in the 802.11 network of Teledyne FLIR Aware2 versions through 6.9.0.2 allows adjacent unauthenticated attackers to intercept, hijack, or modify session traffic against Teledyne FLIR PackBot robots running this software via sniffing or hijacking network traffic.

Unvalidated pathnames in the web interface in Teledyne FLIR Aware2 versions through 6.9.0.2 (PackBot) and 1.7.9 (FirstLook) allows remote unauthenticated attackers to read configuration and security parameters on Teledyne FLIR PackBot and FirstLook robots running this software via path traversal.

Cleartext transmission in the primary control endpoints of Teledyne FLIR Aware2 versions through 6.9.0.2 allows remote unauthenticated attackers to intercept, hijack, or modify session traffic against Teledyne FLIR PackBot robots running this software via sniffing or hijacking network traffic.

Missing authentication in the web interface in Teledyne FLIR Aware2 versions through 6.9.0.2 allows remote unauthenticated attackers to achieve denial of service against Teledyne FLIR PackBot robots running this software via misuse of the reboot endpoint.

Document Merge Service is a document template merge service providing an API to manage templates and merge them with given data. Prior to version 9.1.0, a remote code execution (RCE) via server-side template injection (SSTI) allows for user supplied code to be executed in the server's context where it is executed as the document-merge-server user with the UID 901 thus giving an attacker considerable control over the container. The vulnerability is limited to XLSX templates, were the xltpl library uses a npn-sandboxed Jinja environment for the processing of the template. This issue has been patched in version 9.1.0.

OpenRun is an open-source, self-hosted GitOps platform for deploying web apps and internal tools to Docker or Kubernetes. Prior to version 0.17.7, the restrictions on redirect URLs in openrun can be bypassed by attackers, leading to open redirect attacks. This issue has been patched in version 0.17.7.

NetBox Device Type Library is a collection of community-sourced device type definitions for import into NetBox. Prior to commit f41fc1e, the CI workflow .github/workflows/validation.yml runs on pull_request and executes code supplied by the pull request before any maintainer review. Three PR-editable files drive this: "requirements.txt", ".pre-commit-hooks-config.yaml" / ".pre-commit-yamlfmt-config.yaml", and ".gitmodules". A contributor with no special repository access could open a pull request that modifies these files and have their code run on the CI runner. This issue has been patched via commit f41fc1e.

IVFFlat index build in pgvector before 0.8.7 allows a database user to write data out-of-bounds, which can lead to arbitrary code execution.

The Cadmos LTI application hosted at cadmos.eummena.io had Laravel debug mode enabled (APP_DEBUG=true, APP_ENV=local) in a publicly accessible environment. An unauthenticated attacker could send a GET request and trigger an unhandled exception, causing Laravel to expose the entire server environment, including all .env configuration variables, in plaintext. Fixed on or before 2026-09-02.

Wormhole.app as deployed before 2026-08-22 misconfigures the coturn TURN server and does not properly restrict TCP relay peers, allowing an unauthenticated attacker to access instance metadata or to source TCP connections from the Wormhole relay's IP.

A component of one of the Motorola system applications was exported without permission, allowing for the revocation of runtime permissions from other apps.

The Joyland AI app accepts invalid SSL certificates in the invisible advertisement WebView by default.

Joyland AI app explicitly permits cleartext HTTP traffic on Android 9+ where the default is to block it.

Joyland AI app does not verify hostnames, allowing a malicious host to connect or intercept chat messages.

Sakai is a Collaboration and Learning Environment (CLE). From versions 23.0 to before 23.5, and versions 25.0 to before 25.3, the Sakai Conversations tool stores topic and post messages without HTML sanitization, and the frontend renders them using LitElement's unsafeHTML() directive, resulting in stored cross-site scripting (XSS). Any authenticated user with access to a site that has the Conversations tool enabled can inject arbitrary HTML and JavaScript that executes in the browsers of all other users who view that topic or post. This issue has been patched in versions 23.5, 25.3, and 26.0.

The Joyland AI app accepts any TLS certificates from any server without validation.

Joyland AI app allows an attacker with shared network access to inject JavaScript into content loaded in WebView. Without user-granted permissions, an attacker could access the clipboard, make arbitrary HTTP requests via the Weex 'stream' module, or access app-internal storage. If the installed app has been granted permissions previously, the attacker can access the entire file system, camera, microphone, and GPS tracking.

RT-Labs AB C-Open CANopen contains a write protection bypass in the SDO (Service Data Object) server implementation 'src/co_sdo_server.c' that fails to properly validate write permissions when processing download-segment frames. An unauthenticated attacker on the CAN bus can initiate an SDO upload for a read-only Object Dictionary (OD) entry, which sets a data pointer to the read-only object, then send download-segment frames to write to that memory location. The download-segment handler does not verify that a download session is active, allowing any CANopen node to overwrite read-only OD entries using two SDO frames. Note that CANopen protocol operates over CAN bus and does not provide built-in authentication mechanisms. Fixed in 1.1.1.

The Joyland AI app contains hard-coded credentials for the GeTui push notification service, allowing an attacker to access the GeTui REST API and send push notifications containing arbitrary content to any user, group of users, or all users of the app at once.

RT-Labs AB C-Open CANopen contains a NULL pointer dereference if the LSS protocol is used to configure the device. An object defined by the user application may not have all required subindexes for object 0x1018. An unauthenticated, remote attacker with access to the CAN bus, through a compromised node for instance, can initiate the LSS protocol on a device with a misconfigured identity object and potentially crash the device. Fixed in 1.1.1.

GetSimple CMS is a content management system (CMS), and GetSimple CMS CE is the community edition of that CMS. In versions 3.3.22 and prior, GetSimpleCMS-CE is vulnerable to stored Cross-Site Scripting (XSS) in the "Theme to Components" functionality (admin/components.php) via the title parameter. The stored title is rendered inside a double-quoted HTML attribute in the administrative interface through an output path that HTML-entity-decodes the value before printing it, without re-encoding for the attribute context. This allows persistent execution of arbitrary JavaScript in the admin panel. At time of publication, there are no publicly available patches.