/

CVE-2026-72899

Severity:

CRITICAL RISK

(10.0)

SQL Injection (CWE-89) in Metabase Public Dashboards

CVE-2026-72899

Received

Vulnerability type

Category:

SQL Injection

CWE:

CWE-89

Impact:

Unauthorized Database Access and Data Manipulation

Severity:

CRITICAL RISK

(10.0)

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

A critical SQL injection vulnerability in Metabase may allow an unauthenticated attacker to influence database queries through a publicly shared card or dashboard that exposes a field-filter or dimension parameter. Successful exploitation could expose, modify, or destroy data depending on the permissions of the database connection used by the affected query.

Affected context

The issue may occur in Metabase deployments where cards or dashboards are publicly shared and expose field-filter parameters. Risk is increased when those parameters are incorporated into generated SQL without sufficiently strict validation, type enforcement, or query-safe binding, and when the connected database account has broad privileges.

Exploit preconditions

  • Attack surface: Publicly accessible Metabase cards or dashboards
  • Required access: No Metabase authentication may be required when the affected content is publicly shared
  • Trigger point: Processing a user-controlled field-filter or dimension parameter
  • Why exploitation is possible: Filter input can influence database query construction in a way that may allow SQL syntax or expressions to be interpreted by the backend database

Why this happens

The vulnerability is caused by insufficient separation between user-controlled filter values and SQL query structure. When field-filter or dimension parameters are incorporated into generated queries without robust parameterization, allowlisting, and type-aware validation, crafted input may alter the intended query semantics. This behavior maps to CWE-89: Improper Neutralization of Special Elements used in an SQL Command ('SQL Injection').

Potential impact

  • An attacker may read sensitive records available to the database connection
  • An attacker could bypass intended dashboard filtering and access broader datasets
  • An attacker may alter or delete database records if the connection permits those operations
  • An attacker could obtain database-specific information or metadata
  • Application availability could be affected through resource-intensive queries

General mitigation guidance

Mitigation generally focuses on ensuring that externally supplied filter values cannot become executable SQL structure.

Common defensive patterns include:

  • Using parameterized queries and trusted query-builder abstractions
  • Applying strict allowlists for fields, operators, and permitted filter dimensions
  • Enforcing type-aware validation and rejecting unexpected input before query generation
  • Limiting public sharing of dashboards that expose dynamic database filters
  • Using read-only database accounts with narrowly scoped permissions for analytics workloads
  • Separating public reporting data from sensitive operational databases where practical
  • Applying rate limits and resource controls to publicly accessible query endpoints

How teams detect issues like this

Organizations can look for SQL injection risks through:

  • Static analysis rules identifying string concatenation or interpolation in SQL construction
  • Code review checks for dynamic field, operator, or filter handling
  • Tests that submit unexpected SQL metacharacters and verify that query structure remains unchanged
  • Monitoring for unusual queries, schema enumeration, unexpected data volume, or repeated requests against public dashboards
  • Policy enforcement requiring parameterized database access and restricted database identities
  • Automated security review tools that flag unparameterized query paths and publicly exposed data endpoints

Sources

SourceLink
github.comgithub.com/metabase/metabase/security/advisories/GHSA-r8h2-qpfx-mx59
raw.githubusercontent.comraw.githubusercontent.com/cisagov/CSAF/develop/csaf_files/IT/white/2026/va-26-222-01.json
cve.orgcve.org/CVERecord?id=CVE-2026-72899

Frequently asked

questions

Yes. The described attack surface involves publicly shared cards or dashboards, so an attacker may not need a Metabase account when the affected content is accessible anonymously.

The attacker generally needs only access to a publicly shared dashboard or card that exposes the affected field-filter or dimension parameter. The ultimate impact depends on the permissions of the associated database connection.

Keep user-controlled values separate from SQL syntax through parameterized queries, strict allowlists, and type-aware validation. Public dashboards should expose only narrowly scoped, read-only data sources.

Review all dynamic query-generation paths, test public reporting features with adversarial input, minimize database privileges, and monitor anonymous query activity for abnormal behavior.

Find out what's already
exploitable in your codebase.