A logic error in the legacy WebRTC input handling of the Neko server allows authenticated room members to bypass authorization checks and send mouse and keyboard inputs even when controls are locked. This issue only affects the legacy WebRTC input path when legacy mode and implicit hosting are enabled. The modern control API remains unaffected. The vulnerability is fixed in version 0.0.0-20260805142851-395ca1a6f62b.
Atlantis contains an OS command injection vulnerability in its handling of the Terraform workspace name specified in a pull request's atlantis.yaml file. The workspace name is passed into a shell command without proper escaping, allowing an attacker controlling a pull request to execute arbitrary shell commands as the Atlantis service account during the plan phase. This does not require custom workflows or apply permissions. The vulnerability affects versions from 0.4.0 up to but not including 0.47.1 and has been fixed in version 0.47.1.
FileBrowser contains a vulnerability where symbolic links (symlinks) inside a user's accessible directory can point outside the intended scope, allowing arbitrary file reads outside the served directory. This occurs because the application resolves symlinks without verifying that the resolved target remains within the user's authorized directory. Both authenticated users with download access and unauthenticated users accessing public shares can exploit this to read arbitrary files accessible to the server process. The issue affects version prior to 0.0.0-20260814233832-4126edeb07a0 and is fixed in that version.
CVE-2026-57195 is a high-severity vulnerability in Asterisk's Manager Interface (AMI) that allows an unauthenticated attacker to cause a denial of service via a NULL pointer dereference. This occurs when Digest Authentication is used over HTTP and the HTTP server, AMI, and AMI HTTP access are all enabled. The HTTP server binds to 127.0.0.1 by default, limiting exposure. The issue is fixed in Asterisk versions 20.20.1, 21.12.3, 22.10.1, 23.4.1, and Certified Asterisk 20.7-cert11 and 22.8-cert3.
SkyW4lker is a phishing kit active since June 2026 that targets Microsoft 365 authentication by exploiting OAuth 2.0 device-code flows to bypass MFA. It uses password-protected email attachments leading victims through CAPTCHA pages to malicious sites requesting device codes. The kit primarily targets US firms, especially in technology, manufacturing, and government sectors, and shares infrastructure with the Gabagool phishing kit.
Access of resource using incompatible type ('type confusion') in Microsoft Edge (Chromium-based) allows an unauthorized attacker to execute code over a network.
CVE-2026-116775 is a low-severity vulnerability in the Eugeny russh library affecting versions from 0.2.0 up to but not including 0.2.5. The issue arises from the PageantStream type trusting a peer-controlled length prefix without enforcing the fixed transport capacity limit, which can lead to excessive memory allocation or retention of partial frames. This can cause client memory exhaustion when bridging an SSH forwarded-agent channel to the raw stream. The vulnerability is fixed in version 0.2.5.
CVE-2026-116774 is a resource consumption vulnerability in the russh Rust SSH client and server library versions 0.60.0 through 0.64.9. The flaw allows an authenticated client or malicious server to cause asynchronous write operations to hang indefinitely by specifying a maximum_packet_size of zero, leading to stranded writes and accumulation of affected tasks, channels, and connections without error. This can result in denial of service due to uncontrolled resource consumption. The issue is fixed in version 0.65.0.
CVE-2026-116773 is a resource exhaustion vulnerability in the russh Rust SSH client and server library prior to version 0.65.0. The client improperly handles repeated private EXT_INFO messages (msg::EXT_INFO = 7) sent before authentication, appending extension names to a vector without limits or deduplication. This can lead to progressive memory exhaustion on the client side. The issue is fixed in russh version 0.65.0.
A vulnerability in dnspython's dns.name.from_wire_parser() function can cause excessive CPU usage when processing maliciously crafted DNS messages. This can degrade or deny service to applications that handle untrusted DNS wire format messages using dnspython. The issue is fixed in dnspython version 2.9.0. No workarounds are currently known.
CVE-2026-116696 is a resource allocation vulnerability in the russh Rust SSH client and server library prior to version 0.65.0. The issue occurs in the SSH agent protocol implementation where repeated SSH_AGENT_CONSTRAIN_CONFIRM records can be added without limits, causing unbounded memory growth. This memory growth persists for the lifetime of the process because keys without lifetime constraints do not expire. The vulnerability is fixed in russh version 0.65.0.
Crass, a Ruby CSS parser, has a vulnerability prior to version 1.0.8 where it incorrectly treats a closing parenthesis after an escaped backslash as escaped when parsing bad URLs. This causes Crass to retain CSS declarations inside a bad URL token that browsers would normally treat as live CSS. This can allow CSS-based UI redress, external resource loading, or data exfiltration when sanitizing untrusted CSS, but it does not enable script execution in modern browsers. The issue is fixed in version 1.0.8.
A resource allocation vulnerability exists in open-telemetry's opentelemetry-python prior to version 1.45.0. The JaegerPropagator.extract() function processes inbound uberctx-* baggage values without limits on entry count or byte size. This can lead to excessive CPU usage and memory allocations when untrusted input contains many or large values, potentially causing denial of service. The issue is fixed in version 1.45.0.
CVE-2026-116693 is a high-severity vulnerability in DataDog's dd-trace-js Node.js APM client. Versions from 2.0.0 up to but not including 5.131.1, and from 6.0.0 up to but not including 6.20.1, contain an inefficient regular expression in the default DD_TRACE_OBFUSCATION_QUERY_STRING_REGEXP. This flaw allows crafted URL query strings to cause excessive CPU consumption and stall request processing, potentially making the instrumented application unavailable. The issue is fixed in versions 5.131.1 and 6.20.1.
MLflow versions from 3.12.0rc0 up to but not including 3.17.0 contain a critical code injection vulnerability. This flaw allows an attacker to execute arbitrary Python code by placing a malicious module within a writable allowed package and invoking it via MLflow's artifact endpoints. The vulnerability arises because the deserialization process accepts subclasses of allowed modules rather than strictly enforcing exact module matches. Authentication-disabled deployments are fully exposed, while basic-auth deployments require elevated privileges. The issue is resolved in version 3.17.0.
The dsh-vm-tester npm package versions 0.1.2, 0.1.3, and 0.1.4 contain malicious code that can fully compromise any computer on which it is installed or running. This compromise includes potential unauthorized access to all secrets and keys stored on the affected system. Removal of the package alone does not guarantee elimination of all malicious components, as the attacker may have established persistent control.
MLflow versions from 2.19.0 up to but not including 3.17.0 contain a vulnerability where the mlflow.crewai.autolog() function records CrewAI Agent objects without properly sanitizing nested credential values. This causes provider API keys to be included in plaintext within trace data sent to MLflow's trace backend or OTLP destinations. The issue is resolved in version 3.17.0.
fpdf2 contains a vulnerability where nested SVG <use> references cause exponential growth in memory and CPU usage during image processing. This occurs because fpdf2 deep-copies referenced SVG objects recursively, leading to resource exhaustion. A crafted SVG under 1 KB can cause the process to consume excessive memory and CPU, resulting in a MemoryError and denial of service in PDF generation workflows.
