Overview of incoming advisories.
2186 results found
Paperclip versions before 0.3.1 are vulnerable to remote code execution via DNS rebinding attacks when running in the default local_trusted mode. The vulnerability stems from a failure to validate Host headers, allowing attackers to bypass security controls. An attacker can craft a malicious webpage that, when visited by a developer running Paperclip locally, leverages DNS rebinding to make authenticated API requests. These requests can then execute arbitrary commands through Paperclip's process adapter. The attack targets developers running Paperclip in local development environments. The fix was introduced in version 0.3.1. This type of attack is particularly dangerous as it exploits the trusted local environment assumption. Developers should upgrade to version 0.3.1 or later immediately to mitigate the risk.
View original advisory →CVE-2026-77264 describes a critical authentication bypass vulnerability in the Automation Web Platform – Notifications and OTP for WooCommerce, Advanced Country Code plugin for WordPress, affecting versions up to and including 4.8.6. The vulnerability exists in the handle_email_otp_return() function, which incorrectly returns the secret magic login token directly in the HTTP response to a publicly accessible OTP request, instead of delivering it exclusively to the user's email. This design flaw allows unauthenticated attackers to capture the token and use it to log in as any registered user on the site, including administrators, as long as they know the target's email address. The attack requires no prior authentication and can lead to full site compromise. Site owners using the affected plugin versions are urged to update immediately to a patched version.
View original advisory →CVE-2026-54457 affects TensorZero, an open-source LLMOps platform, prior to version 2026.6.0. The vulnerability exists in the /internal/object_storage endpoint, which accepts a caller-supplied JSON storage_path parameter that can override the object storage configuration. Attackers can exploit the filesystem storage type to read arbitrary files from the gateway filesystem, including sensitive credential files. The s3_compatible storage type can be abused to trigger server-side request forgery (SSRF) against internal or cloud metadata endpoints such as AWS IMDS. Exploitation depends on whether the gateway is deployed with or without authentication, making the attack surface variable. The vulnerability combines arbitrary file read and SSRF primitives, posing significant risk in cloud-hosted deployments. A fix was released in version 2026.6.0. Patches and advisories are available via the official GitHub repository.
View original advisory →CVE-2026-77236 is a vulnerability in FreeRTOS-Kernel versions prior to 11.3.1 involving missing minimum size validation in secure context allocation. The flaw may allow local users to corrupt secure-world heap metadata through an out-of-bounds write triggered by supplying an undersized stack size parameter. This affects the secure context allocation mechanism, potentially impacting systems relying on FreeRTOS TrustZone or secure-world isolation. The vulnerability is classified as a local privilege or integrity issue, as it requires local access to exploit. AWS has published a security bulletin acknowledging the issue. The fix is available in FreeRTOS-Kernel version 11.3.1 and later. Users and organizations using affected versions are strongly advised to upgrade immediately to remediate the risk.
View original advisory →CVE-2026-48752 affects Incus, a system container and virtual machine manager. Prior to version 7.2.0, a specially crafted image or instance backup can be leveraged to read or create/write arbitrary files on the host system. This path traversal-style vulnerability could potentially lead to arbitrary command execution on the host. The vulnerability is exploitable via malicious container images or instance backups. Version 7.2.0 has been released to patch the issue. Users are advised to upgrade to version 7.2.0 or later to mitigate the risk. The severity is high given the potential for host-level compromise from within a container context.
View original advisory →CVE-2026-62940 affects Incus, a system container and virtual machine manager, in versions prior to 7.3.0. The vulnerability exists in the instance migration process between cluster members, where user-supplied configuration overrides are applied without enforcing project restriction policies. Security-critical configuration keys such as `security.privileged` and `raw.lxc` can be manipulated by restricted project users during migration. This allows an attacker to escalate privileges to a privileged container and subsequently escape to the underlying host system. The flaw represents a container escape vulnerability with significant impact on multi-tenant cluster environments. Version 7.3.0 of Incus addresses and patches this issue. Users running Incus in clustered configurations with restricted project users are at elevated risk.
View original advisory →Combodo iTop, a web-based IT service management tool, contains a vulnerability prior to version 3.2.3 where inline images accessible without authentication are protected only by a weak 24-bit pseudo-random secret. This weak randomness could allow an attacker to brute-force or predict the secret and gain unauthorized access to inline images. The vulnerability affects all versions of iTop prior to 3.2.3. The issue has been patched in version 3.2.3. The fix is available via a commit on the official GitHub repository. No authentication is required to exploit this vulnerability, increasing its risk surface. Organizations using iTop for IT service management should upgrade to version 3.2.3 immediately. The vulnerability is tracked as CVE-2026-27490 and has a GitHub Security Advisory associated with it.
View original advisory →BabelDOC, a document translation tool, contains a critical vulnerability in its vendored PDF parser (babeldoc/pdfminer/cmapdb.py) prior to version 0.6.3. The vulnerability allows arbitrary Python code execution via unsafe deserialization of pickle data when CMapDB._load_data() loads CMap files. Attackers can craft malicious PDF files with controlled Encoding or CMapName values, or embed PostScript usecmap operators, to manipulate file paths. The _normalize_cmap_name() function only strips a leading slash, leaving path traversal sequences intact. Combined with os.path.join() behavior, absolute paths or traversal sequences can escape trusted CMap directories and load attacker-controlled .pickle.gz files. This triggers pickle.loads() to execute arbitrary Python code with the privileges of the BabelDOC process. The vulnerability is fixed in version 0.6.3.
View original advisory →UAC (Unix-like Artifacts Collector) versions prior to 3.3.0 contain a command injection vulnerability in the parse_artifact.sh script. The flaw exists in the user substitution logic where usernames and home directories read from /etc/passwd are directly interpolated into command strings without proper escaping before being executed via eval. An attacker who can control /etc/passwd entries (e.g., on a compromised host being analyzed) can craft malicious usernames or home directory paths containing shell metacharacters such as command substitution syntax or semicolons. When an analyst runs UAC against such a system, the injected commands execute on the analyst's own host system. This represents a significant risk to forensic investigators and incident responders using UAC. The vulnerability has been patched in UAC version 3.3.0.
View original advisory →The internment crate version 0.8.7 for the Rust programming language has been identified as containing a rogue dependency that executes malicious code at compile time. When a developer compiles a project that depends on this crate, the malicious dependency silently registers with a command-and-control (C2) server, enabling arbitrary code execution on the developer's machine. This is classified as a build-time supply chain attack, where the compromise occurs during the software build process rather than at runtime. The attack vector is linked to a broader supply chain campaign involving the arrayref and proc-macro1 Rust crates. Multiple security advisories and blog posts from the Rust community, RustSec, SafeDep, and StepSecurity have documented this incident. The vulnerability is tracked as CVE-2026-77649 and RUSTSEC-2026-0266. Developers using the internment crate 0.8.7 are urged to update or remove the dependency immediately.
View original advisory →Combodo iTop, a web-based IT service management tool, contains a Reflected Cross-Site Scripting (XSS) vulnerability in its synchro import script. The vulnerability is identified as CVE-2026-30890 and affects all versions prior to 3.2.3. Reflected XSS vulnerabilities allow attackers to inject malicious scripts into web pages viewed by other users, potentially leading to session hijacking, credential theft, or other client-side attacks. The issue has been patched and fully remediated in iTop version 3.2.3. The fix is available via a specific commit on the official Combodo iTop GitHub repository. A security advisory has also been published on GitHub under GHSA-93q9-fc8m-5gp5. Users of iTop are strongly advised to upgrade to version 3.2.3 or later to mitigate this risk.
View original advisory →Combodo iTop, a web-based IT service management tool, contains a reflected Cross-Site Scripting (XSS) vulnerability prior to version 3.2.3. The vulnerability exists in the dashboard revert functionality, specifically through the 'dashboard_id' parameter in the '/pages/ajax.render.php' endpoint. An attacker could exploit this flaw to inject malicious scripts that execute in the context of a victim's browser. The issue has been assigned CVE-2026-30819 and is documented in the NVD. A fix has been released in iTop version 3.2.3. The patch is available via a GitHub commit and a corresponding security advisory has been published on GitHub. Users are strongly advised to upgrade to version 3.2.3 or later to mitigate the risk.
View original advisory →CVE-2026-77811 describes a stored Cross-Site Scripting (XSS) vulnerability in the dashboards-observability plugin for OpenSearch Dashboards. The flaw stems from improper input validation, allowing a remote authenticated user with write permissions to saved objects to upload malicious assets containing arbitrary web content. This enables the attacker to execute arbitrary JavaScript in the browser sessions of other users viewing the affected content. The attack requires authentication and write access to OpenSearch Dashboards saved objects, making it a privilege-dependent but high-impact issue. Successful exploitation could lead to session hijacking, credential theft, or further lateral movement within the platform. AWS has issued a security bulletin and patches are available in OpenSearch releases 2.19.6 and 3.4.0. Organizations using affected versions of OpenSearch Dashboards should upgrade immediately to mitigate the risk.
View original advisory →GNU Emacs versions before 31.0.91 contain an integer overflow vulnerability in the PBM/PPM/PGM image loader. The flaw arises from signed integer arithmetic used when multiplying image dimensions and channel counts, which can wrap to a negative value for sufficiently large inputs. This negative result bypasses bounds checks, allowing the pixel reader to access heap memory beyond the allocated buffer. An attacker can exploit this by supplying a crafted image with large dimensions and an elevated max color index. The over-read heap memory contents are then interpreted as pixel color values and rendered on screen, effectively leaking sensitive heap data. A fix is available in GNU Emacs 31.0.91.
View original advisory →CVE-2026-62283 affects Nezha Monitoring, a self-hostable server and website monitoring tool. Affected versions include 1.14.13 through 1.14.14 and 2.0.0 through 2.0.9. The vulnerability stems from stream identifiers created by CreateStream in service/rpc/io_stream.go not being bound to their creating user. The WebSocket endpoints GET /ws/terminal/:id and GET /ws/file/:id only verify UUID existence without validating ownership. An authenticated RoleMember who obtains a valid stream UUID can hijack another user's terminal or file-manager session. This allows the attacker to read and write files on the target server and execute arbitrary shell commands. The UUID could be obtained via logs, browser history, referer headers, or telemetry data. The vulnerability is fixed in version 2.0.10, with a patch available in the referenced GitHub commit.
View original advisory →A critical arbitrary code execution vulnerability exists in the Hydra configuration framework prior to version 1.3.4. The flaw resides in hydra.utils.instantiate(), which resolves and calls Python objects via _resolve_target() based on configuration values. An attacker who can control configuration inputs, CLI overrides, or model metadata can force the instantiation of dangerous Python callables, leading to arbitrary code execution within the application process. Potential impacts include reading or modifying files, credential theft, and process termination. The vulnerability affects any application, library, CLI workflow, or model loader that passes untrusted input into hydra.utils.instantiate(). Version 1.3.4 mitigates the issue by introducing target blocking with an allowlist mechanism controlled by the HYDRA_INSTANTIATE_ALLOWLIST_OVERRIDE environment variable. Users are strongly advised to upgrade to version 1.3.4 immediately.
View original advisory →CVE-2026-55622 affects Incus, a system container and virtual machine manager, in versions prior to 7.2.0. The vulnerability stems from missing authorization checks during instance copying operations. An attacker who knows the name of a project and an instance within that project—even without authorized access—can copy the instance to a different project. This unauthorized copying could expose sensitive secrets stored within those instances. The flaw represents a significant access control bypass in multi-tenant or shared container environments. Version 7.2.0 of Incus has been released to patch this issue. Users are strongly advised to upgrade to mitigate unauthorized access risks.
View original advisory →CVE-2026-49114 affects ONNX versions before 1.21.0, where the 'save_external_data' function constructs external-data file paths without safe file-opening flags (O_NOFOLLOW/O_EXCL). A non-atomic os.path.isfile() check creates a TOCTOU race condition. A local attacker with write access to the serialization directory can pre-plant a symlink before the victim writes data. This allows the attacker to redirect writes to arbitrary files the victim can write, such as ~/.ssh/authorized_keys, cron files, or application configs. The vulnerability enables privilege escalation or persistent access through file manipulation. It is classified as a local symlink-follow attack. The issue has been fixed in ONNX version 1.21.0.
View original advisory →CVE-2026-61539 affects Xinference versions 2.5.0 and earlier, an inference API for open-source and multimodal models. The vulnerability arises from passing attacker-influenced Llama3 tool-call output directly to Python's eval() function in llama3_tool_parser.py and utils.py. Requests to the /v1/chat/completions endpoint with a tools field traverse multiple processing layers before the unsafe eval() call is made. An unauthenticated remote attacker can craft a malicious prompt to influence the model-generated Python expression passed to eval(). This allows arbitrary command execution within the Xinference server process context, representing a critical remote code execution risk. No authentication is required to exploit this vulnerability, significantly broadening the attack surface. The flaw was fixed in Xinference version 2.7.0 via a patch committed to the xorbitsai/inference GitHub repository.
View original advisory →A denial-of-service vulnerability exists in the kin-openapi Go library (versions 0.10.0 through 0.140.x) due to a nil pointer dereference in openapi3filter/validation_error_encoder.go. The function convertParseError dereferences e.Parameter.In without first verifying that e.Parameter is non-nil. A malformed non-string scalar field in a multipart/form-data request body triggers a nested ParseError with a nil RequestError.Parameter, causing applications using openapi3filter.ConvertErrors or ValidationErrorEncoder to panic. An unauthenticated remote attacker can repeatedly send such crafted requests to exhaust service availability when no recovery boundary (e.g., panic recovery middleware) is in place. JSON request bodies and applications not using the affected error-rendering helpers are not impacted. The vulnerability is fixed in version 0.141.0 of kin-openapi.
View original advisory →