| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| llama.cpp is an inference of several LLM models in C/C++. Integer Overflow in the gguf_init_from_file_impl function in ggml/src/gguf.cpp can lead to Heap Out-of-Bounds Read/Write. This vulnerability is fixed in commit 26a48ad699d50b6268900062661bd22f3e792579. |
| pdfme is a TypeScript-based PDF generator and React-based UI. The expression evaluation feature in pdfme 5.2.0 to 5.4.0 contains critical vulnerabilities allowing sandbox escape leading to XSS and prototype pollution attacks. This vulnerability is fixed in 5.4.1. |
| The web-push crate before 0.10.3 for Rust allows a denial of service (memory consumption) in the built-in clients via a large integer in a Content-Length header. |
| Zipkin through 3.5.1 has a /heapdump endpoint (associated with the use of Spring Boot Actuator), a similar issue to CVE-2025-48927. |
| arduino-esp32 provides an Arduino core for the ESP32. Versions prior to 3.3.0-RC1 and 3.2.1 contain a HTTP Response Splitting vulnerability. The `sendHeader` function takes arbitrary input for the HTTP header name and value, concatenates them into an HTTP header line, and appends this to the outgoing HTTP response headers. There is no validation or sanitization of the `name` or `value` parameters before they are included in the HTTP response. If an attacker can control the input to `sendHeader` (either directly or indirectly), they could inject carriage return (`\r`) or line feed (`\n`) characters into either the header name or value. This could allow the attacker to inject additional headers, manipulate the structure of the HTTP response, potentially inject an entire new HTTP response (HTTP Response Splitting), and/or ause header confusion or other HTTP protocol attacks. Versions 3.3.0-RC1 and 3.2.1 contain a fix for the issue. |
| jackson-core contains core low-level incremental ("streaming") parser and generator abstractions used by Jackson Data Processor. In versions prior to 2.15.0, if a user parses an input file and it has deeply nested data, Jackson could end up throwing a StackoverflowError if the depth is particularly large. jackson-core 2.15.0 contains a configurable limit for how deep Jackson will traverse in an input document, defaulting to an allowable depth of 1000. jackson-core will throw a StreamConstraintsException if the limit is reached. jackson-databind also benefits from this change because it uses jackson-core to parse JSON inputs. As a workaround, users should avoid parsing input files from untrusted sources. |
| Integer Overflow or Wraparound vulnerability in dragonflydb dragonfly (src/redis/lua/struct modules). This vulnerability is associated with program files lua_struct.C.
This issue affects dragonfly: 1.30.1, 1.30.0, 1.28.18. |
| Out-of-bounds Read vulnerability in dail8859 NotepadNext (src/lua/src modules). This vulnerability is associated with program files lparser.C.
This issue affects NotepadNext: through v0.11.
The singlevar() in lparser.c lacks a certain luaK_exp2anyregup call, leading to a heap-based buffer over-read that might affect a system that compiles untrusted Lua code. |
| RISC Zero is a zero-knowledge verifiable general computing platform, with Ethereum integration. The risc0-ethereum repository contains Solidity verifier contracts, Steel EVM view call library, and supporting code. Prior to versions 2.1.1 and 2.2.0, the `Steel.validateCommitment` Solidity library function will return `true` for a crafted commitment with a digest value of zero. This violates the semantics of `validateCommitment`, as this does not commitment to a block that is in the current chain. Because the digest is zero, it does not correspond to any block and there exist no known openings. As a result, this commitment will never be produced by a correct zkVM guest using Steel and leveraging this bug to compromise the soundness of a program using Steel would require a separate bug or misuse of the Steel library, which is expected to be used to validate the root of state opening proofs. A fix has been released as part of `risc0-ethereum` 2.1.1 and 2.2.0. Users for the `Steel` Solidity library versions 2.1.0 or earlier should ensure they are using `Steel.validateCommitment` in tandem with zkVM proof verification of a Steel program, as shown in the ERC-20 counter example, and documentation. This is the correct usage of Steel, and users following this pattern are not at risk, and do not need to take action. Users not verifying a zkVM proof of a Steel program should update their application to do so, as this is incorrect usage of Steel. |
| Meshtastic-Android is an Android application for the mesh radio software Meshtastic. Prior to version 2.5.21, an attacker is able to send an unencrypted direct message to a victim impersonating any other node of the mesh. This message will be displayed in the same chat that the victim normally communicates with the other node and it will appear as using PKC, while it is not. This means that the victim will be provided with a false sense of security due to the green padlock displayed when using PKC and they'll read the attacker's message as legitimate. Version 2.5.21 contains a patch for the issue. It is suggested to implement a stricter control on whether a message has been received using PKC or using the shared Meshtastic channel key. Moreover, instead of showing no green padlock icon in the chat with no PKC, consider using an explicit indicator like, for example, the yellow half-open padlock displayed when in HAM mode. This remediation, however, applies to the client applications rather than the Meshtastic firmware. |
| Claude Code is an agentic coding tool. Claude Code extensions in VSCode and forks (e.g., Cursor, Windsurf, and VSCodium) and JetBrains IDEs (e.g., IntelliJ, Pycharm, and Android Studio) are vulnerable to unauthorized websocket connections from an attacker when visiting attacker-controlled webpages. Claude Code for VSCode IDE extensions versions 0.2.116 through 1.0.23 are vulnerable. For Jetbrains IDE plugins, Claude Code [beta] versions 0.1.1 through 0.1.8 are vulnerable. In VSCode (and forks), exploitation would allow an attacker to read arbitrary files, see the list of files open in the IDE, get selection and diagnostics events from the IDE, or execute code in limited situations where a user has an open Jupyter Notebook and accepts a malicious prompt. In JetBrains IDEs, an attacker could get selection events, a list of open files, and a list of syntax errors. Claude released a patch for this issue on June 13th, 2025. Although Claude Code auto-updates when a user launch it and auto-updates the extensions, users should take the following steps, though the exact steps depend on one's integrated development environment (IDE). For VSCode, Cursor, Windsurf, VSCodium, and other VSCode forks, check the extension Claude Code for VSCode. Open the list of Extensions (View->Extensions), look for Claude Code for VSCode among installed extensions, update or uninstall any version prior to 1.0.24, and restart the IDE. For JetBrains IDEs including IntelliJ, PyCharm, and Android Studio, check the plugin Claude Code [Beta]. Open the Plugins list, look for Claude Code [Beta] among installed extensions, update or uninstall any version prior to 0.1.9, and restart the IDE. |
| CSV Injection vulnerability in Addactis IBNRS v.3.10.3.107 allows a remote attacker to execute arbitrary code via a crafted .ibnrs file to the Project Description, Identifiers, Custom Triangle Name (inside Input Triangles) and Yield Curve Name parameters. |
| xmedcon 0.23.0 and fixed in v.0.24.0 is vulnerable to Buffer Overflow via libs/dicom/basic.c which allows an attacker to execute arbitrary code. |
| SummaryThis advisory addresses a security vulnerability in Mautic where unpublished page previews could be accessed by unauthenticated users and potentially indexed by search engines. This could lead to the unintended disclosure of draft content or sensitive information.
Unauthorized Access to Unpublished Page Previews: The page preview functionality for unpublished content, accessible via predictable URLs (e.g., /page/preview/1, /page/preview/2), lacked proper authorization checks. This allowed any unauthenticated user to view content that was not yet intended for public release, and allowed search engines to index these private preview URLs, making the content publicly discoverable.
MitigationMautic has patched this vulnerability by enforcing proper permission checks on preview pages. Users should upgrade to the patched version of Mautic or later. |
| Mail-0's Zero is an open-source email solution. In version 0.8 it's possible for an attacker to craft an email that executes javascript leading to session hijacking due to improper sanitization. This issue has been patched in version 0.81. |
| Improper Prevention of Lock Bit Modification in SEV firmware could allow a privileged attacker to downgrade firmware potentially resulting in a loss of integrity. |
| Improper bound check within AMD CPU microcode can allow a malicious guest to write to host memory, potentially resulting in loss of integrity. |
| Improper Access Control in an on-chip debug interface could allow a privileged attacker to enable a debug interface and potentially compromise data confidentiality or integrity. |
| CycloneDX Sunshine v0.9 is vulnerable to CSV Formula Injection via a crafted JSON file |
| Archer 6.11.00204.10014 allows attackers to execute arbitrary code via crafted system inputs that would be exported into the CSV and be executed after the user opened the file with compatible applications. NOTE: the Supplier does not accept this as a valid vulnerability report against their product. |
