stigmem-node EUVD-2026-62477

The stigmem-plugin-multi-tenant opt-in plugin must be installed and active on the node - this is a non-default configuration. The attacker must hold a valid tenant administrator account with the tenant write capability on at least one tenant hosted on the target node; node-level superadmin authority is not required and is not a limiting factor. Exploitation is confined to the /v1/quarantine endpoints (list, count, get, admit, reject); no other endpoint surfaces are reported as affected. Default single-tenant nodes are completely unaffected because no second tenant exists to cross into. Access is network-based and requires valid credentials; unauthenticated or anonymous exploitation is not possible.

The CVSS 4.0 score of 8.6 (AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N) reflects a network-reachable, low-complexity flaw requiring only authenticated tenant-level access with no availability impact. The high confidentiality and integrity scores are appropriate: an attacker can exfiltrate quarantined content and alter moderation decisions (admit or reject facts) across all tenants on the node. However, the real-world risk is sharply bounded by a mandatory deployment prerequisite - the stigmem-plugin-multi-tenant plugin must be opted in; default single-tenant nodes are confirmed unaffected. No EPSS score was provided in the input data and no active exploitation has been reported. The risk profile is high for affected multi-tenant deployments and negligible for everyone else. No conflicting signals in available data; CVSS vector is consistent with the described attack.

A tenant administrator on tenant-B of a multi-tenant stigmem node sends an authenticated GET request to /v1/quarantine/list; because the list query lacks a tenant_id predicate, the response includes quarantined facts belonging to tenant-A. The attacker then issues a POST to /v1/quarantine/{fact_id}/admit using a fact ID from tenant-A, successfully admitting that tenant's content using only their tenant-B write credential. No public exploit code is known at time of analysis.