Elasticsearch CVE-2026-56144

Exploitation requires a valid Elasticsearch account with at least limited index-level privileges (PR:L per CVSS); anonymous or unauthenticated access is insufficient. The target index must have an ingest pipeline configured - indices without pipelines will not yield pipeline-processed data output, though index mapping metadata disclosure may still be achievable regardless of pipeline presence. Attack complexity is High (AC:H) because the attacker cannot control whether a target index has pipelines configured; this is a deployment-dependent condition. The vulnerability does not provide write access, affect availability, or allow scope change to other systems. Exploitation is most impactful in multi-tenant Elasticsearch clusters where strict index-level role separation is enforced and users hold credentials with partial access only.

The vendor-assigned CVSS score of 5.3 (Medium) reflects a constrained but genuine confidentiality risk. The CVSS vector AV:N/AC:H/PR:L/UI:N/S:U/C:H/I:N/A:N identifies several limiting factors: low-privilege authenticated access is required (PR:L), and attack complexity is rated High (AC:H), consistent with the precondition that target indices must have ingest pipelines configured - a specific deployment condition rather than a universal default. The confidentiality impact is rated High (C:H) because pipeline output can expose sensitive processed data, not merely raw field values. No EPSS data was provided, no CISA KEV listing exists, and no public proof-of-concept code has been identified, all of which reduce urgency relative to a KEV-listed disclosure. The S:U scope confirms the exploit cannot pivot to other systems. In multi-tenant or shared-cluster Elasticsearch deployments - where multiple user roles coexist with granular index-level permissions - this vulnerability carries meaningful real-world risk because the trust model is violated: index isolation is a core security assumption of role-based access control in Elasticsearch. In single-tenant or fully trusted environments where all authenticated users are effectively equivalent, impact is negligible.

An attacker holding a valid Elasticsearch account with read access to a limited set of low-sensitivity indices sends an ingest simulation request that specifies a restricted, high-value index (such as one storing enriched PII or financial records) as the target. The simulation API executes the target index's configured ingest pipeline - which may perform data lookups, field extraction, or enrichment - and returns the full processed output to the attacker. The attacker iterates across index names (potentially enumerated via the mapping metadata disclosure path) to harvest pipeline outputs and schema information for indices entirely outside their authorized scope.