vLLM versions prior to 0.32.0 contain a resource exhaustion vulnerability in the P2P secondary tier management. An API client can submit many unique peer addresses that never establish connections, causing vLLM to retain sessions and ZeroMQ sockets indefinitely. This leads to exhaustion of the ZeroMQ context socket quota, raising a ZMQError that terminates the EngineCore and shuts down the API server. The vulnerability allows denial of service without requiring privileged access or internal listener connectivity. A fix is available in vLLM 0.32.0.
fpdf2 versions prior to 2.8.8 implicitly load external HTTP(S) URLs and local filesystem image paths when rendering content via its image and HTML/SVG rendering APIs. This can lead to outbound network requests or local file access attempts if untrusted input controls image references. The vulnerability may enable Server-Side Request Forgery (SSRF) and limited local file access depending on the content. A fix is available in version 2.8.8.
MONAILabel server is vulnerable to a denial-of-service and silent inference corruption attack via crafted DICOM files exploiting pydicom's improper validation of Decimal String (DS) and Integer String (IS) values. An unauthenticated attacker can upload malicious DICOM files containing NaN or infinity values in spatial metadata fields, causing the server to crash or produce corrupted AI inference results. This affects MONAILabel versions prior to 1.6.1 and pydicom versions 0.9.7 through 3.0.2. The vulnerability arises because MONAILabel trusts pydicom output without validating these invalid values, and the REST API accepts uploads without authentication by default. A fix is available in MONAILabel 1.6.1.
Ajenti versions prior to 2.2.13 contain a critical remote code execution vulnerability (CVE-2026-27975) that allows unauthenticated users to execute arbitrary code on the affected server. This vulnerability has a CVSS score of 9.8, indicating a high severity with network attack vector, no privileges required, and no user interaction needed. The issue is fixed in version 2.2.13, and users are strongly advised to upgrade to this version promptly to mitigate the risk.
CVE-2026-117573 is an OS command injection vulnerability in hagent version 0. It allows MCP clients to execute arbitrary shell commands by injecting special characters into the args parameter of the hagent_arduino and hagent_esp32 tools. The vulnerability arises because these args are interpolated into arduino-cli and idf.py commands executed with subprocess.run(shell=True), enabling command execution with the privileges of the MCP server user.
CVE-2026-117569 is an OS command injection vulnerability in the kali-gpt project, specifically in the execute_tool() function within kali_gpt/tools/mcp_server.py. This flaw allows MCP clients to execute arbitrary OS commands by injecting shell metacharacters into unsanitized parameters such as target, ports, domain, wordlist, or url. The vulnerability arises because these parameters are passed to subprocess.run with shell=True, running commands as the MCP server user. The CVSS 4.0 base score is 8.5, indicating high severity.
YesWiki versions prior to 4.7.0 have a server-side request forgery (SSRF) vulnerability in the Bazar CSV import function copyUrlToLocalFile(). Authenticated administrators can exploit this by supplying crafted URLs in CSV or image fields, enabling arbitrary outbound requests. The vulnerability allows attackers to bypass extension checks using query-string suffixes or redirects to access cloud metadata credentials, internal services, or local files via file:// URLs. Responses are stored in the web-accessible files/ directory. This issue is fixed in version 4.7.0.
Multiple vulnerabilities affecting AMD EPYC server CPUs based on Zen 2 (Rome), Zen 3 (Milan), and Zen 4 (Genoa) core generations were disclosed by AMD in November 2023. Google has applied fixes to its infrastructure, including Google Cloud and Compute Engine, to protect customers. No evidence of exploitation has been reported. Customers using Google Cloud services do not need to take any action as mitigations are already in place.
A vulnerability in crossbario autobahn-python versions 26.7.1 through 26.10.0 allows an unauthenticated WebSocket client to cause a memory exhaustion denial of service. The issue arises because the Brotli decompressor is called without an output limit, allowing a small, highly compressed frame to expand to hundreds of MiB before the server enforces the maximum message payload size. This flaw affects servers that enable the permessage-brotli feature via the compress extra. The vulnerability is fixed in version 26.10.1.
Tornado versions prior to 6.6.0 have a vulnerability in HTTP/1.1 chunked transfer encoding decoding where the process event loop can be monopolized by a remote unauthenticated attacker sending many one-byte chunks. This stalls other connections and can cause high memory consumption due to chunk retention. The issue affects both server handlers and the SimpleAsyncHTTPClient. The vulnerability is fixed in Tornado 6.6.0.
vLLM versions from 0.16.0 up to but not including 0.32.0 contain a vulnerability in the /v1/responses endpoint. An authenticated user can supply a crafted template in chat_template_kwargs[chat_template] when no server template is configured. This bypasses the intended trust policy and may cause CPU or memory exhaustion due to Jinja template rendering. The issue is resolved in version 0.32.0.
vLLM versions prior to 0.32.0 have a misconfiguration in the OpenAI-compatible server where Starlette CORSMiddleware allows all origins by default. This allows a malicious web page to make cross-origin requests from a victim's browser to an unauthenticated vLLM server on localhost, internal networks, or VPNs, potentially consuming GPU resources and reading responses. The vulnerability does not enable cookie-based CSRF attacks as credentials are disabled by default, and bearer token authentication limits exposure to already authorized callers. The issue is fixed in version 0.32.0.
vLLM versions from 0.19.0 up to but not including 0.32.0 contain a vulnerability where authenticated users can supply a request chat template if no server-side template is configured. This allows a malicious template to be compiled and rendered without execution limits, potentially causing denial of service by blocking response handling or exhausting server memory. Harmony models are not affected. The issue is fixed in version 0.32.0.
Interpretation Conflict vulnerability in hackney_url in benoitc hackney allows Server Side Request Forgery via a percent-encoded bracketed IPv6 literal or an IP literal with a trailing dot. A host such as %5B%3A%3Affff%3A127%2E0%2E0%2E1%5D (decoding to [::ffff:127.0.0.1]) or 127.0.0.1. is not an IP address to a caller's allowlist built on uri_string:parse/1 and inet:parse_address/1, while hackney connects to the address it denotes. The same applies to a Location target when redirects are followed.
The IP-literal gate that hackney_url:normalize/2 applies to a percent-decoded host, added to close the percent-encoded IP bypass of CVE-2026-47076, passes the decoded value to inet_parse:address/1 as is. The connect path in hackney_happy:parse_address/1 drops a surrounding bracket pair, and the resolver ignores a trailing dot, so both forms pass the gate and still reach the literal address. Domain names with a trailing dot and unencoded IPv6 literals are not affected.
This issue affects hackney: from 4.0.1 before 4.9.0.
Frappe Builder versions before 1.36.1 have a server-side request forgery (SSRF) vulnerability in the builder/api.py component. The vulnerability arises because the assert_not_private_url() function only validates the initial hostname of a URL, but does not handle HTTP redirects properly, allowing attackers to redirect requests to internal or link-local addresses. An authenticated attacker can exploit this via the convert_to_webp endpoint, which lacks permission checks, to perform SSRF attacks, internal port scanning, or cause denial of service by triggering request hangs due to missing timeouts. The issue is fixed in version 1.36.1.
A use-after-free vulnerability exists in rustls-ffi versions 0.12.0 through 0.15.6 where client connection constructors retain a reference to the caller's server_name buffer instead of copying it. If the caller frees or reuses this buffer after the constructor returns, TLS processing may read freed memory. This affects clients that do not keep the server_name buffer alive for the lifetime of the rustls_client_config. The issue is fixed in version 0.15.7 and later.
A security update for python3.9 on Red Hat Enterprise Linux 9 addresses multiple vulnerabilities including an arbitrary file deletion via race condition during temporary directory cleanup (CVE-2026-12345), a certificate verification bypass due to missing server_hostname validation in SSLContext.wrap_bio() (CVE-2026-19553), and a use-after-free vulnerability in server-side SSLContext when sni_callback switches contexts (CVE-2026-19445). These issues have been rated with an important security impact by Red Hat Product Security.
Multiple security vulnerabilities have been identified and fixed in the OpenSSH implementation on Red Hat Enterprise Linux 9.2. These include local man-in-the-middle attacks on X11 forwarding, client-side denial of service due to double free, sftp client file location control, use-after-free during host key re-exchange, sshd security bypass related to forwarding and tunneling, and scp file misplacement during remote copy. These issues affect both the OpenSSH client and server components.