scim-patch EUVD-2026-85327

Requires an actor able to submit SCIM PATCH operations to an application that uses scim-patch prior to 0.9.2; the attacker supplies a PATCH 'op' whose path (e.g. 'toString.scimPatchPolluted', 'valueOf.x', or a nested/array-search variant like 'emails[primary eq true].toString.x') or a dotted value key begins with an inherited built-in property. This is authenticated (PR:L) - not exploitable by a fully anonymous attacker unless the SCIM endpoint accepts unauthenticated PATCH input. No special non-default configuration or user interaction is needed (AC:L, UI:N), but meaningful real-world impact additionally requires that the application later reads the polluted inherited-method property in a security-relevant code path; absent that, the pollution is largely inert.

This is a genuine-but-modest issue, and the signals are internally consistent rather than conflicting. The vendor CVSS of 4.3 (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N) matches the described mechanism: scim-patch's navigate() reads inherited properties and assign() uses prototype-chain 'in' membership checks, so an actor who can submit SCIM PATCH operations can steer a path or dotted value key through an inherited built-in (e.g. toString) and add attacker-controlled properties, mutating shared process-global function objects (CWE-915 prototype pollution). Impact is integrity-only and limited (I:L, C:N, A:N) because the pollution lands on built-in method objects and only affects application logic that later reads inherited-method properties - it is not a direct RCE, DoS, or data-exfiltration primitive on its own. No EPSS score, KEV listing, or SSVC decision was supplied, and there is no weaponized public exploit - only the maintainer's own regression tests in the fix PR demonstrate the write. PR:L means exploitation requires the ability to send PATCH operations (an authenticated/provisioning-capable actor), which meaningfully limits blast radius. Bottom line: patch it during normal maintenance cadence, but this is a low-to-moderate real-world priority, not an emergency - its severity depends heavily on whether the consuming application reads polluted inherited properties in a security-sensitive path.

An attacker with the ability to send SCIM PATCH requests (e.g. a provisioning client or a tenant in a multi-tenant identity integration) submits an operation whose path or dotted value key begins with an inherited property such as 'toString.scimPatchPolluted', causing scim-patch to traverse into the shared Object.prototype.toString function and attach an attacker-controlled property that persists process-wide. No public exploit code exists; only the vendor's fix-PR regression tests demonstrate the write, and exploitation complexity is low (AC:L) but requires PATCH-submission privileges (PR:L).