/

CVE-2026-92796

Severity:

HIGH RISK

(8.8)

Improper Authorization (CWE-863) in Manticore Search SQL

CVE-2026-92796

Deferred

Vulnerability type

Category:

Improper Authorization

CWE:

CWE-863

Impact:

Unauthorized data access and potential administrative account compromise

Severity:

HIGH RISK

(8.8)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Manticore Search does not consistently enforce permissions across every statement in a multi-statement SQL request, allowing read-only users to perform unauthorized queries. An attacker who can submit such a request may read credential data, obtain password hashes, and potentially authenticate as an administrator without recovering plaintext passwords.

Affected context

The issue occurs in Manticore Search deployments that permit multi-statement SQL requests while applying authorization checks only to the initial statement. It becomes risky when a lower-privileged account can submit additional statements and the database exposes credential or other sensitive tables to the affected query path.

Exploit preconditions

  • Attack surface: SQL query interface that accepts multi-statement requests
  • Required access: An authenticated account with at least read-only database access
  • Trigger point: Processing additional statements appended to an otherwise permitted request
  • Why exploitation is possible: Authorization checks are not consistently applied to every statement in the request
  • Additional condition: Sensitive credential records must be readable through the accessible query context

Why this happens

The underlying bug is an incomplete authorization model for batched or multi-statement SQL requests. The system validates the permission for one statement but fails to independently authorize subsequent statements, allowing a lower-privileged principal to perform operations outside its intended scope. This behavior maps to CWE-863, Improper Authorization, because access-control decisions do not consistently enforce the user's granted permissions.

Potential impact

  • Attackers may read credential tables or other restricted data
  • Password hashes could be disclosed
  • Compromised hashes may enable authentication as administrative users
  • Unauthorized users could gain broader access to search data and management functions
  • Sensitive information may be exposed or used for follow-on account compromise

General mitigation guidance

Mitigation generally focuses on making authorization decisions statement-aware

and enforcing least privilege throughout the query-processing pipeline.

Common defensive patterns include:

  • Authorizing every statement independently rather than trusting the first statement
  • Disabling multi-statement requests for roles that do not require them
  • Applying strict allowlists for permitted SQL operations and accessible objects
  • Restricting direct access to credential and system tables
  • Storing authentication material with appropriate protection and minimizing its query exposure
  • Using separate, least-privileged accounts for read-only workloads
  • Enforcing authentication, network boundaries, and administrative access controls

How teams detect issues like this

Organizations can look for this weakness through:

  • Static analysis and code review of authorization logic around SQL batching and statement splitting
  • Tests that submit mixed-permission multi-statement requests under each database role
  • Policy checks confirming that sensitive tables are inaccessible to read-only principals
  • Query auditing for unexpected access to credential tables or administrative objects
  • Monitoring for authentication activity involving accounts or credentials outside normal usage patterns
  • Automated security testing that verifies authorization is applied consistently to every parsed statement

Sources

SourceLink
github.comgithub.com/geo-chen/oss/blob/main/manticoresearch.md
github.comgithub.com/manticoresoftware/manticoresearch
github.comgithub.com/manticoresoftware/manticoresearch/blob/release-28.4.4/src/auth/auth_proto_mysql.cpp#L408-L418
github.comgithub.com/manticoresoftware/manticoresearch/issues/4706
vulncheck.comvulncheck.com/advisories/manticore-search-27.0.0-before-28.4.4-multi-statement-authorization-bypass

Frequently asked

questions

It may be exploitable remotely when the Manticore SQL interface is network-accessible

and an attacker has valid credentials for a lower-privileged account. Exposure depends

on network controls and how the service is configured.

The attacker generally needs an authenticated account that can submit SQL requests,

such as a read-only user. Anonymous access is not inherently required.

Enforce authorization independently for every statement in a multi-statement request,

and disable multi-statement execution where it is not necessary.

Use least-privilege roles, restrict access to credential tables, test every query

execution path with role-specific authorization cases, and monitor for anomalous

sensitive-data access.

Find out what's already
exploitable in your codebase.