CVE-2026-112085
webpack-dev-server is a development server that provides live reloading for webpack projects. In versions before 5.2.7 and in version 6.0.0,
CVSS
8.2
Alta
EPSS
0.1%
p4
KEV
—
Exploit Today
1
0-100
Publicado: 22 de out. de 2026 · Última mod.: 29 de out. de 2026 · CWE-346 · CWE-1385
0.1%EPSS · 30 dias0.2%
2026-10-222026-11-06
webpack-dev-server is a development server that provides live reloading for webpack projects. In versions before 5.2.7 and in version 6.0.0, when a proxy entry is configured to forward WebSockets, the server forwards non-HMR WebSocket upgrade requests to the proxy target without validating the Host or Origin headers, so the allowedHosts protection does not apply to those proxied upgrades. A malicious web page visited by a developer can open a cross-origin WebSocket through the dev server and read from and write to the proxied backend, even when the server is bound to loopback. The issue is fixed in webpack-dev-server 5.2.7 and 6.1.0, where proxied WebSocket upgrades pass the same Host and Origin validation as the HMR WebSocket, and users should upgrade to one of those versions or later. As a workaround, omit WebSocket forwarding from proxy entries that do not need it and scope the proxy to specific paths.
CVECVSSEPSSKEVRExploitTítuloMod.
CVE-2026-1023448.1 ALT—
———Lavish is the new editor for your HTML artifacts. Prior to 0.1.43, the lavish-axi loopback HTTP server did not authenticate requests or validate Host and Origin values on session, artifact, prompt, and polling routes. A website visited while the server is running can use DNS rebinding to send same-origin browser requests to the local service. POST /api/sessions accepts a caller-selected file path and returns its deterministic session key; GET /artifact/:key/index.html then returns the file contents, allowing disclosure of any file readable by the server process without an existing session. An attacker can also queue text through POST /api/:key/prompts for an active agent session when the artifact path is known or guessed. Visiting the attacker-controlled website is the required user interaction. This vulnerability is fixed in 0.1.43.6hCVE-2026-1166305.9 MÉD—
——0undici is an HTTP client for Node.js. In versions 7.0.0 through 7.30.1 and 8.0.0 through 8.12.3, when an Agent composes the redirect and shared cache interceptors, a cached response that is revalidated with a conditional request can be validated by a 304 Not Modified from a different origin. When the origin that stored the response redirects the conditional revalidation request to another origin, the redirect handler strips origin-bound credential headers but not the cache-generated If-None-Match or If-Modified-Since validators, so the redirect destination receives the original origin's validator and undici treats its 304 as successful revalidation of the stored representation. As a result, a shared cache can serve the original origin's stored authenticated response, including one gated by Authorization, to a later unauthenticated caller. The issue is fixed in undici 7.30.2 and 8.12.4, and users should upgrade to one of those versions or later. As a workaround, compose the interceptors as cache then redirect, use per-origin dispatchers and cache stores, or disable shared caching on redirect-capable requests.7hCVE-2026-1165927.4 ALT—
——0undici is an HTTP client for Node.js. In versions 7.30.1 and 8.12.2 through 8.12.3, when an originless Agent composes the redirect and shared cache interceptors, undici decides whether to store the final response under the original request key by inspecting only the final redirect hop, which bypasses the earlier fix for CVE-2026-109829 that compared only the initial and final origins. A redirect that bounces through an untrusted origin and returns to the trusted origin is still cached under the trusted key, an intermediate hop's Vary is dropped so a response that varies on Authorization or Cookie can be served to a different user, and an intermediate hop's no-cache or default-uncacheable status is ignored so a non-storable redirect is laundered into a cacheable entry. An attacker who controls a redirect target from a trusted URL can poison the cached entry for that URL or have a private, per-user response served to a later caller. The issue is fixed in undici 7.30.2 and 8.12.4, and users should upgrade to one of those versions or later. As a workaround, compose the interceptors as cache then redirect, use per-origin dispatchers and cache stores, or disable shared caching on redirect-capable requests.7hCVE-2026-1232728.0 ALT9.4%
——3Mark3labs mcp-go before v1.2.1 contains an improper validation vulnerability in OAuthHandler.fetchServerMetadata that allows malicious MCP servers to obtain tokens for other resources by declaring arbitrary resources in well-known metadata. Attackers can declare a different MCP server as the RFC 8707 resource indicator, causing the authorization server to mint tokens the attacker replays against that resource.1dCVE-2026-1232718.0 ALT3.9%
——1ChatCLI before v1.215.2 contains an improper validation vulnerability in the MCP OAuth login flow that uses server-supplied Protected Resource Metadata as the resource indicator without checking it matches the configured server URL. A malicious MCP server can declare another resource on the same authorization server, obtain the user's bearer token for it, and replay that token against the other resource.1dCVE-2026-119344—23.2%
——7Cocos AI is a confidential computing system for AI. In version 0.8.2 and earlier, Cocos AI accepts trusted execution environment attestation Evidence when firmware patch levels and software measurements match, without binding the Evidence to an allowed platform owner or unique machine identity and without enforcing revocation. A remote attacker who controls or compromises another platform trusted by the same verifier can substitute valid Evidence from that platform and cause authorization of the wrong platform. This vulnerability is addressed in 0.9.0. However, this vulnerability can't be patched in source code alone, since the vulnerability arises from the confidential computing threat model itself.1d