Satori EUVD-2026-89994

Exploitation requires that the application pass attacker-controlled values into SVG content, attributes, or styles during image generation - concretely, the Node.js ImageResponse implementation from next/og (Next.js >=16.2.0 <16.3.6) or direct use of Satori (0.0.27-0.33.4) to render untrusted input. Limiting factors: the Edge ImageResponse implementation is NOT affected; applications that never route attacker-controlled data into SVG are NOT affected (this AT:P deployment prerequisite, not authentication, is the main brake on exploitation); no login is needed once such a route exists (PR:N). The only complete remediation is upgrading (Satori 0.33.5 / Next.js 16.3.6); the interim mitigation is to stop rendering attacker-controlled content through the Node.js path.

This is a genuine priority for the affected population despite the modest 5.3 CVSS 4.0 base score, because that vendor score was computed against Satori-the-library in isolation (VC:N/VI:N/VA:N with subsequent-system SC:H/SI:H) whereas the real-world consumer, Next.js next/og, escalates the identical flaw to remote code execution per the Next.js GHSA-vcvr-r3jv-pc5j advisory. The CVSS 4.0 vector (AV:N/AC:L/AT:P/PR:N/UI:P) shows an unauthenticated, network-reachable issue whose only friction is a deployment prerequisite (AT:P): the application must feed attacker-controlled data into SVG content/attributes/styles. No EPSS score, KEV listing, or public weaponized exploit was provided, so there is no evidence of active exploitation yet; however the advisories include a concrete reproduction pattern that makes exploit development straightforward. Signals conflict: the numeric CVSS reads 'medium' but the vendor-described impact is RCE - the divergence is an artifact of scoping the score to the library rather than the consuming runtime, and defenders should treat next/og Node.js deployments as high-risk. Note the tag 'Information Disclosure' is inconsistent with the CWE-116 injection-to-RCE nature of this flaw.

A Next.js application exposes an OG-image route that inserts a user-supplied query parameter into SVG (e.g. inside a <title> or attribute) via the Node.js next/og ImageResponse. An unauthenticated attacker sends a crafted request whose value breaks out of the intended text context and injects SVG markup, which - because Satori fails to escape it - is interpreted as active markup and, in the Node.js rendering path, escalates to remote code execution on the server. No public weaponized exploit was provided, but the vendor advisories include a working reproduction pattern.