Skip to main content
CVE Vulnerability Database
Vulnerability Database/CVE-2026-63738

CVE-2026-63738: SurrealDB Auth Bypass Vulnerability

CVE-2026-63738 is an authentication bypass flaw in SurrealDB versions 3.1.0 to 3.1.4 that allows attackers to bypass field-level permissions through graph traversals. This article covers technical details, impact, and mitigation.

Published:

CVE-2026-63738 Overview

CVE-2026-63738 is a broken access control vulnerability [CWE-863] in SurrealDB, a multi-model database supporting graph, document, and relational workloads. Versions from 3.1.0 up to but not including 3.1.5 fail to enforce field-level SELECT permissions when records are materialized through graph-edge or back-reference traversals. An authenticated user with table-level SELECT access can retrieve field values that field-level permissions were configured to hide. The vulnerability affects confidentiality only and requires low privileges to exploit over the network.

Critical Impact

Authenticated users can bypass field-level SELECT permissions by traversing graph edges or back-references, exposing protected field values in SurrealDB deployments running versions 3.1.0 through 3.1.4.

Affected Products

  • SurrealDB 3.1.0
  • SurrealDB versions 3.1.1 through 3.1.4
  • Fixed in SurrealDB 3.1.5

Discovery Timeline

  • 2026-07-20 - CVE-2026-63738 published to NVD
  • 2026-07-22 - Last updated in NVD database

Technical Details for CVE-2026-63738

Vulnerability Analysis

SurrealDB supports layered permissions where administrators grant table-level SELECT access and then restrict individual fields with field-level permission clauses. This model assumes the field-level checks apply consistently regardless of how a record is reached. The vulnerability breaks that assumption. When a record is materialized through a graph traversal or back-reference rather than a direct table scan, the query planner skips the field-level authorization step and returns the underlying field values to the caller.

The flaw undermines the confidentiality guarantees of least-privilege schema design in multi-tenant applications built on SurrealDB. Sensitive attributes such as password hashes, personally identifiable information, financial fields, or internal identifiers can be surfaced by any user who already holds table-level read access on a related table.

Root Cause

The root cause is an incomplete authorization path in the query execution layer. Field-level SELECT permissions are evaluated during direct record access but are not applied when records are produced as the target of a graph edge traversal (for example, ->edge->target) or through back-reference expressions. The permission check is coupled to the access path rather than to the record and field being returned, which is the pattern identified by CWE-863 (Incorrect Authorization).

Attack Vector

An attacker authenticates to the SurrealDB instance with any account that has legitimate table-level SELECT access. Instead of issuing a direct SELECT query against the protected table, the attacker issues a query that reaches the target records through a graph traversal or a back-reference from a related table. Fields that would normally be filtered by field-level permissions are returned in the result set. Exploitation requires only network access to the database endpoint and valid low-privilege credentials, and it produces no memory corruption or availability impact. Refer to the GitHub Security Advisory GHSA-hv6h-hc26-q48p and the VulnCheck Advisory for technical details.

Detection Methods for CVE-2026-63738

Indicators of Compromise

  • Query logs containing graph-edge traversals (->, <-) or back-reference expressions that target tables holding fields protected by PERMISSIONS FOR select clauses.
  • Result sets returned to low-privilege sessions that include fields those sessions cannot retrieve through direct table SELECT queries.
  • Repeated traversal queries from a single session enumerating related records across sensitive tables.

Detection Strategies

  • Enable SurrealDB query auditing and compare fields returned via traversal queries against the fields each role is authorized to read directly.
  • Baseline expected traversal patterns per application role and alert on deviations that reach tables containing hidden fields.
  • Correlate authentication events with query patterns to identify accounts issuing traversal queries they have not previously used.

Monitoring Recommendations

  • Ship SurrealDB query and audit logs to a centralized analytics platform for long-term retention and cross-session correlation.
  • Track the SurrealDB server version in asset inventories and flag any host still running 3.1.0 through 3.1.4.
  • Monitor egress volume from the database tier for unexpected spikes that may indicate bulk enumeration through traversals.

How to Mitigate CVE-2026-63738

Immediate Actions Required

  • Upgrade SurrealDB to version 3.1.5 or later on all affected instances.
  • Inventory schemas that rely on field-level SELECT permissions and review recent query logs for traversal-based access to those fields.
  • Rotate credentials or secrets stored in fields that were protected by field-level permissions and may have been exposed.

Patch Information

SurrealDB 3.1.5 resolves the issue by enforcing field-level SELECT permissions on records materialized through graph traversals and back-references. The fix is documented in the SurrealDB Security Advisory GHSA-hv6h-hc26-q48p. Apply the upgrade following SurrealDB's standard release procedure and validate that dependent applications continue to function after the permission model is tightened.

Workarounds

  • Restrict database accounts so that users who must not read hidden fields also lack SELECT access on tables that are reachable through graph edges pointing to those records.
  • Remove or redesign graph edges and back-references that terminate on tables containing fields protected by field-level SELECT permissions until the upgrade is applied.
  • Place the SurrealDB endpoint behind network controls that limit access to trusted application tiers, reducing the number of accounts that could abuse the traversal path.

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Default Legacy - Prefooter | Experience the World’s Most Advanced Cybersecurity Platform

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.