Overview of incoming advisories.
1553 results found
CVE-2026-54789 affects mod_auth_openidc, an OpenID Certified authentication and authorization module for Apache 2.x HTTP servers. The vulnerability exists in the state-cookie parser, where an out-of-bounds read and a one-byte out-of-bounds write can be triggered by malformed Cookie headers containing tokens that lack an '=' character. The flaw impacts all versions prior to 2.4.19.4. The fix was introduced in version 2.4.19.4 by enforcing a scan stop at the string terminator, causing value-less tokens to be rejected. No in-product workarounds are available, making upgrading the primary recommended remediation. As a temporary mitigation, deploying an upstream reverse proxy or WAF that rejects or normalizes malformed Cookie headers can reduce exposure. The vulnerability poses a risk to systems relying on OpenID Connect authentication via Apache modules.
View original advisory →CVE-2026-77767 describes an improper access control vulnerability in Reconmap's API where the report preview endpoint in ReportsController.cs is decorated with [AllowAnonymous], bypassing the global administrator-only authorization policy. The PreviewReport action loads project and client organisation data using a sequential auto-increment integer ID without any authentication, membership, or role check. An unauthenticated remote attacker can enumerate all project IDs sequentially and extract sensitive penetration-testing engagement details including project names, descriptions, and client organisation information such as name, address, and URL. The 404 response for missing IDs further aids enumeration by revealing which project IDs exist. Since Reconmap is used to manage penetration-testing engagements, the exposed data is highly sensitive by nature. A fix is available via a specific commit to the repository.
View original advisory →LeafWiki versions 0.1.0 through 0.10.0 contain a privilege escalation vulnerability in the user update API. An authenticated user can modify their own account role, escalating from a regular role such as 'viewer' to 'admin'. Exploitation only requires a valid authenticated account on the affected instance. Instances with restricted registration and only trusted users face lower practical risk. The vulnerability has been patched in version 0.10.1. As a mitigation, operators should restrict account creation and limit access to the user update API to trusted users or administrators only. The issue is tracked as CVE-2026-53527 and a security advisory has been published on GitHub.
View original advisory →CVE-2026-62941 affects Incus, a system container and virtual machine manager, in versions prior to 7.3.0. The vulnerability exists in the cross-project instance copy workflow, where the project restriction check (AllowInstanceCreation) runs before the source instance configuration is merged into the request. This timing flaw allows dangerous configuration keys such as security.privileged, raw.lxc, and raw.apparmor to be injected into the target project after the security check has already passed. As a result, all project-level restrictions on the target project are effectively bypassed. An attacker with access to copy instances across projects could exploit this to escalate privileges or escape container isolation. The issue has been patched in Incus version 7.3.0. Users are advised to upgrade immediately to mitigate the risk.
View original advisory →CVE-2026-77815 affects the Infinite Image Browsing extension for Stable Diffusion WebUI. The vulnerability stems from the use of os.path.normpath in the to_abs_path function, which collapses dot segments but does not resolve symbolic links. An attacker can place a symlink inside a scanned directory that points to sensitive files outside the directory, such as /etc/passwd, bypassing the is_path_trusted containment check. The FileResponse handler then follows the symlink and serves the targeted file, resulting in arbitrary file disclosure. The flaw is most impactful in network-exposed deployments where access control is active (e.g., when using share, ngrok, listen, or server_name options). The fix replaces os.path.normpath with os.path.realpath to properly resolve symbolic links before path comparison.
View original advisory →CVE-2026-62676 affects Omnigent, an open-source AI agent framework, prior to version 0.3.0. A flawed shell-command parser in omnigent/policies/builtins/_shell.py fails to recognize combined interpreter flags, timeout/nice/setsid/stdbuf wrappers, command substitutions, and background control operators. This allows malicious or prompt-injected commands to bypass policy enforcement, specifically the github.py write_repos and write_branches allowlist and the working_dir.py workspace confinement policies. An authenticated or prompt-injected agent can exploit this to push to unauthorized repositories or branches, or escape the intended workspace sandbox. The vulnerability stems from the parser abstaining (allowing) commands it cannot parse rather than defaulting to deny. The issue has been patched in Omnigent version 0.3.0. Users are advised to upgrade immediately to prevent unauthorized repository access or workspace escape.
View original advisory →CVE-2026-54682 is a stored Cross-Site Scripting (XSS) vulnerability in DiscordChatExporter prior to version 2.47.2. When HTML exports are generated with markdown formatting disabled, attacker-controlled content is passed through FormatMarkdownAsync and FormatEmbedMarkdownAsync without HTML entity encoding. Affected fields include message content, forwarded/referenced message content, embed titles, descriptions, and field names/values. A malicious Discord webhook or bot can inject script payloads into these fields. The payload executes when a victim exports the channel to HTML and opens the resulting file in a browser. This allows the attacker to read export contents or alter displayed content. The vulnerability is fixed in DiscordChatExporter version 2.47.2.
View original advisory →CVE-2026-77775 affects Headroom's LLM proxy, which allows clients to specify an arbitrary upstream destination via the x-headroom-base-url request header. The proxy functions _resolve_openai_upstream_base and _select_passthrough_base_url accept this header without validating the destination, permitting requests to loopback, link-local, and RFC 1918 (private) addresses, enabling Server-Side Request Forgery (SSRF). Internal services and cloud metadata endpoints (e.g., AWS IMDSv1) can be reached and their responses returned to the caller. Additionally, the Authorization header is forwarded unchanged to the attacker-designated host, risking credential leakage. The default pip install binds to 127.0.0.1, but the shipped docker-compose.yml exposes the service on 0.0.0.0 without requiring HEADROOM_PROXY_TOKEN authentication, greatly widening the attack surface. A fix was introduced in v0.36.1 via upstream_guard.py. Deployments following the reference compose configuration are exposed to unauthenticated network-level exploitation.
View original advisory →A path traversal vulnerability exists in Atlantis, a self-hosted Go application for Terraform pull request automation, affecting versions 0.19.8 through 0.44.x. The flaw stems from insufficient validation of user-controlled workspace values provided via repository-level atlantis.yaml configuration or the authenticated /api/plan endpoint. Malicious traversal segments can escape the intended workspace directory, causing filesystem operations such as os.RemoveAll and os.MkdirAll to be executed on unintended directories. These operations run with the full privileges of the Atlantis process, enabling attackers to create, delete, or manipulate arbitrary directories. The impact includes integrity loss and potential denial of service. The vulnerability requires either repository-level access (to modify atlantis.yaml) or authenticated API access, limiting the attack surface but not eliminating risk. The issue has been patched in Atlantis version 0.45.0. Users are advised to upgrade immediately to mitigate the risk.
View original advisory →CVE-2026-77776 affects Headroom's LLM proxy, which incorrectly derives memory ownership from the client-supplied x-headroom-user-id HTTP request header without binding it to an authenticated caller. This allows any client to impersonate another user and read or write that user's stored LLM memory. The vulnerability exists in headroom/proxy/handlers/openai.py across chat completion and websocket paths. The fix introduces a resolve_memory_identity function in headroom/proxy/identity.py that validates the header only for loopback or allowlisted callers. Compounding the issue, the reference docker-compose.yml ships with --host 0.0.0.0 and no required HEADROOM_PROXY_TOKEN, exposing the proxy to the network without authentication. Deployments following the shipped compose configuration are particularly at risk of unauthorized access to user LLM memory data.
View original advisory →A missing privilege verification flaw exists in the secure context cleanup handler of FreeRTOS-Kernel versions prior to 11.3.1. The vulnerability allows local users to trigger a use-after-free condition in secure-world memory by exploiting the SVC handler for secure context deallocation. This affects systems using TrustZone-enabled ARM processors where FreeRTOS manages secure and non-secure contexts. The flaw stems from inadequate privilege checks before performing cleanup operations in secure memory regions. Successful exploitation could lead to memory corruption or potential code execution in the secure world. AWS has issued a security bulletin acknowledging the issue. The fix is available in FreeRTOS-Kernel version 11.3.1 and users are strongly advised to upgrade immediately.
View original advisory →YOURLS, a self-hosted PHP URL shortener, contains a stored XSS vulnerability affecting versions 1.5.1 through 1.10.3. The application stores the HTTP Referer header without proper sanitization and later concatenates derived domain values into inline JavaScript within statistics pages without JavaScript-string escaping. An unauthenticated attacker can craft a malicious Referer header to poison the statistics of any existing short URL. When an administrator or public stats-page viewer opens the affected statistics page, attacker-controlled JavaScript executes within the YOURLS origin. This can expose admin-visible data, the API signature token, and allow privileged same-origin actions. The attack chain involves multiple functions: yourls_get_referrer(), yourls_sanitize_url_safe(), yourls_log_redirect(), yourls_get_domain(), yourls_stats_pie(), and yourls_google_array_to_data_table(). The vulnerability is fixed in YOURLS version 1.10.4.
View original advisory →UAC (Unix-like Artifacts Collector) versions prior to 3.3.0 contain a command injection vulnerability in the _command_collector function. The flaw arises because foreach command output lines are substituted directly into command strings via sed without proper escaping, and then evaluated using eval. Attackers can exploit this by crafting malicious filenames or artifact definitions containing shell metacharacters such as command substitution syntax or semicolons. Successful exploitation allows arbitrary command execution on the analyst's host system. The vulnerability is particularly concerning as it targets forensic analysts who use UAC for incident response and artifact collection. A fix was introduced in UAC version 3.3.0. The patch is available via a GitHub commit and pull request. Security advisory details are also published by VulnCheck.
View original advisory →CVE-2026-55241 affects Checkmate, an open-source self-hosted server monitoring tool, prior to version 3.9.1. The vulnerability exists in the public POST /api/v1/auth/register endpoint, which processes multipart profileImage uploads via in-memory Multer parsing before any registration validation occurs. The upload middleware in server/src/api/middleware/upload.ts lacks file-size, file-count, and MIME-type restrictions. An unauthenticated attacker can exploit this by submitting concurrent oversized file uploads, causing files to be buffered in memory before invalid registration or invite-token checks reject the request. This leads to memory exhaustion, potentially crashing or destabilizing the backend server. The vulnerability is fixed in Checkmate version 3.9.1.
View original advisory →Combodo iTop, a web-based IT service management tool, contains a Reflected Cross-Site Scripting (XSS) vulnerability tracked as CVE-2026-30826. The vulnerability exists in the OQL (Object Query Language) query testing functionality. An attacker could exploit this flaw to inject malicious scripts that are reflected back to users, potentially leading to session hijacking, credential theft, or other client-side attacks. The vulnerability affects all versions of iTop prior to 3.2.3. Combodo has addressed the issue in version 3.2.3. A fix was committed to the official GitHub repository and a security advisory was published. Users are strongly advised to upgrade to version 3.2.3 or later to mitigate the risk.
View original advisory →A command injection vulnerability has been discovered in Comfast CF-N1-S firmware version 2.6.0.1. The flaw resides in the system function accessible via the CGI endpoint /cgi-bin/mbox-config?method=SET§ion=ntp_timezone. Attackers can exploit this by manipulating the 'timestr' argument to inject arbitrary OS commands. The vulnerability is remotely exploitable without requiring physical access to the device. A public exploit has already been released, significantly increasing the risk of active exploitation. The affected product is a network device (likely a wireless access point or router), making this a network infrastructure concern. The CVE has been tracked and documented on NVD, VulDB, and GitHub. Given the public exploit availability and remote exploitability, this vulnerability poses a high risk to organizations using the affected hardware. Users are advised to apply patches or mitigations as soon as they become available.
View original advisory →A critical improper input validation vulnerability (CVE-2026-77234) has been identified in FreeRTOS-Kernel versions prior to 11.3.1. The flaw affects MPU-enabled ports and could allow an unprivileged task to execute arbitrary code in a privileged kernel context, effectively bypassing memory protection unit boundaries. This represents a significant privilege escalation risk in embedded and real-time operating system environments. The vulnerability is particularly concerning given FreeRTOS's widespread use in IoT and embedded systems. AWS has issued a security bulletin (2026-086-aws) acknowledging the issue. The remediation requires upgrading to FreeRTOS-Kernel version 11.3.1 or later, which is available on the official GitHub repository. No workarounds are mentioned; patching is the recommended course of action.
View original advisory →CVE-2026-48753 affects Incus, a system container and virtual machine manager. Prior to version 7.1.0, the S3 protocol upload endpoint is vulnerable to path traversal attacks. This vulnerability allows attackers to create arbitrary files on the host system. The ability to write arbitrary files can lead to arbitrary command execution on the host. The vulnerability is fixed in Incus version 7.1.0. Users are advised to upgrade to version 7.1.0 or later to remediate the issue. The flaw is documented in both the NVD and a GitHub Security Advisory.
View original advisory →The arrayref crate version 0.3.10 for Rust has been identified as containing a rogue dependency that constitutes a supply chain attack. When developers compile a project that includes this crate, malicious code can be triggered during the build process. The rogue dependency registers with a command-and-control (C2) server, enabling arbitrary code execution on the victim's machine at compile time. This is a build-time malware attack, meaning developers do not need to run the software for compromise to occur. The attack was disclosed in August 2026 and has been documented across multiple security advisories including RustSec and the Rust language blog. The incident highlights risks associated with open-source dependency ecosystems and the potential for malicious packages to be introduced into trusted registries. Security researchers from SafeDep and StepSecurity have published analyses of the attack vector. The affected crate is listed in the RustSec advisory database under RUSTSEC-2026-0260.
View original advisory →CVE-2026-62674 affects Omnigent, an open-source AI agent framework for orchestrating coding agents, in versions prior to 0.3.0. The vulnerability exists in the PUT /sessions/{session_id}/agent endpoint, which checks LEVEL_EDIT permissions but fails to reject bound shared or template agents with a null agent.session_id. An authenticated user with edit access to a session can replace a shared agent bundle, inject a malicious stdio MCP server, and cause subsequent sessions using that shared agent to execute attacker-controlled commands. The commands run with Omnigent runner process permissions, potentially exposing files, credentials, workspace data, and internal services. The attack vector requires authentication but allows privilege escalation through the shared agent mechanism. Affected code paths include omnigent/server/routes/sessions.py and omnigent/tools/mcp.py. The vulnerability has been fixed in version 0.3.0 via a patch available on GitHub.
View original advisory →