A vulnerability in the Model Context Protocol (MCP) Python SDK client versions 1.10.0 through 1.29.1 and 2.0.0 through 2.1.1 allows a malicious MCP server to cause the client to synchronously fetch external URLs or local files referenced in tool output schemas. This can lead to client-side network requests or file reads without user consent, and can stall the client process by holding connections open indefinitely. The issue is fixed in versions 1.30.0 and 2.2.0 or later by restricting reference resolution to within the schema document only.
Red Hat has announced a Technology Preview container image for the satellite/foreman-mcp-server-rhel9, designed for advanced reporting and AI-driven data analysis within Red Hat Satellite. Among multiple CVEs associated with this advisory, CVE-2026-102267 is detailed as a significant vulnerability in the PyJWT library used for JSON Web Key Set (JWKS) retrieval. This flaw allows an attacker who can manipulate a trusted JWKS endpoint's HTTP redirect responses to redirect key requests to untrusted servers, potentially leading to acceptance of forged cryptographic keys or exposure of sensitive credentials. No patches or mitigations meeting Red Hat's criteria are currently available for this issue. The advisory does not specify affected product versions or confirm active exploitation in the wild.
The @sylphx/pdf-reader-mcp package, when run as an HTTP server with MCP_TRANSPORT=http, fails to enforce API key authentication despite logging that it does. This allows unauthenticated access to the /mcp endpoint, enabling any client able to reach the server to invoke PDF processing tools and read any accessible PDF files on the host. The default binding to all interfaces (0.0.0.0) increases exposure risk. This vulnerability affects versions >=2.2.0 and <3.0.1 and is fixed in version 3.0.1.
mcp-sanitizer version 1.3.3 contains a Server-Side Request Forgery (SSRF) vulnerability that allows bypassing localhost and IPv6 loopback address restrictions when URLs include explicit ports or use bracketed IPv6 notation. This flaw enables attacker-controlled URLs to reach internal services, potentially exposing sensitive data. The issue arises from improper hostname checks that fail to block localhost URLs with ports and bracketed IPv6 addresses. A fix is available in version 1.4.0.
CVE-2026-61542 is a high-severity vulnerability in drhelius Gearboy emulator version prior to 3.8.7. The Gearsystem component's optional --mcp-http mode accepts unauthenticated MCP requests from any network peer, allowing remote peers to invoke emulator and debugging operations without authentication. The vulnerability arises because the MCP server binds to all interfaces and does not enforce authorization checks on POST requests to /mcp. The impact is limited to the emulator process, debugging state, and local files used as media, with no demonstrated host code execution. Gearboy 3.8.7 includes a patch that fixes this issue.
CVE-2026-59176 affects the functype TypeScript functional programming libraries prior to version 1.4.4. The vulnerability allows an attacker to supply an unconstrained version string to the set_functype_version MCP tool, leading to arbitrary package installation and dynamic code execution within the MCP server process. This can be exploited by any MCP client invoking the tool, which is enabled by default without authentication in stdio mode. In certain deployments, remote network attackers can exploit this without local access. The flaw is fixed in version 1.4.4.
The AI Engine – MCP Server & AI Chatbot for WordPress plugin is affected by a path traversal vulnerability that allows authenticated users with subscriber-level access or higher to delete arbitrary media files on the server. This vulnerability exists due to insufficient file path validation in the upload_file function in versions up to and including 3.8.4. Exploitation requires the Public API option to be enabled by an administrator and depends on the scheduled hourly cleanup_files task to execute after a malicious upload.
CVE-2026-116684 is an authorization vulnerability in the Model Context Protocol Python SDK versions 1.10.0 through before 1.30.0 and 2.0.0 through before 2.2.0. The flaw occurs when server-side bearer authentication accepts a valid token issued for a different service due to lack of independent audience (aud) validation by the TokenVerifier. This allows an attacker holding a token for one resource to authenticate to a different MCP server and gain unauthorized access. The issue is fixed in versions 1.30.0 and 2.2.0.
Certain versions of the MCP TypeScript SDK and related packages do not limit HTTP POST request body size or JSON-RPC batch length. This allows an attacker to send very large requests or long batches, potentially exhausting server memory and causing denial of service. No authentication is required unless implemented outside the SDK. Fixed versions impose limits on request size and batch length.
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.
CVE-2026-117567 is an OS command injection vulnerability in esp-mcp (horw project) present through commit 28afddf. It allows attackers to execute arbitrary OS commands by injecting shell metacharacters into parameters such as target, port, project_path, or project_name used by the MCP tool. This occurs when these parameters are passed unsafely to asyncio.create_subprocess_shell, enabling command execution as the MCP server user.