Description

A missing input-validation issue in MongoDB libmongocrypt's automatic-encryption context setup allows a caller-supplied database identifier to be accepted without sanitization. The resulting impact is limited to incorrect schema selection, which may lead to limited disclosure or modification of information handled by the application.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "4.4",
  "pubDate": "2026-08-27T20:18:50.633Z",
  "pubdate": "2026-08-27T20:18:50.633Z",
  "executiveSummary": "A critical input validation vulnerability exists within the libmongocrypt library, specifically during the automatic-encryption context setup phase. The flaw arises from the library's failure to sanitize caller-supplied database identifiers.\nThis vulnerability allows an attacker to manipulate the context selection process, leading to the application interacting with unintended database schemas. The impact of successful exploitation includes potential information disclosure or unauthorized data modification within the context of the application's database operations.\nThe vulnerability affects systems utilizing libmongocrypt for automatic encryption. The risk is primarily centered on data integrity and confidentiality breaches. Exploitation does not require elevated network privileges but necessitates the ability to influence the configuration parameters passed to the library during the context initialization sequence.\nAttackers can leverage this flaw to force the client-side encryption logic to use an incorrect schema, effectively bypassing security controls that rely on schema-specific encryption policies.",
  "technicalDetails": "The root cause of this vulnerability lies in the insufficient sanitization of input parameters within the libmongocrypt automatic-encryption context setup. Specifically, the library accepts a caller-supplied database identifier without implementing necessary validation or canonicalization routines.\nThe vulnerability manifests during the creation of an encryption context, where the library maps an incoming request to a specific schema. Because the database identifier is processed without sanitization, a malicious caller can inject specific identifiers that cause the library to erroneously select a different schema than the one intended by the application developer.\nThe attack flow follows a sequential process: First, the attacker identifies a code path where the application invokes the libmongocrypt context setup. Second, the attacker manipulates the input parameters, specifically the database identifier field, to reference an unauthorized or alternative database namespace. Third, the library, lacking proper validation, processes this identifier and binds the encryption context to the incorrect target schema.\nOnce the encryption context is incorrectly established, any subsequent cryptographic operations—such as the encryption or decryption of fields—will utilize the security metadata defined by the malicious schema selection. This behavior allows for potential unauthorized disclosure of sensitive data if an attacker can force the use of a weaker or different encryption policy, or facilitate data modification if the misconfigured schema allows for unexpected field mappings.\nThis vulnerability is localized to the libmongocrypt library logic governing automatic encryption metadata resolution. It does not strictly require external network exposure, as the threat model involves manipulating inputs already flowing through the application's internal data processing pipelines. Consequently, the impact is confined to the scope of the application's database interactions and the specific permissions granted to the database user under which the application operates.\nPost-exploitation, an attacker gains the ability to circumvent intended data protection mechanisms by controlling the schema context. This effectively undermines the guarantees provided by Queryable Encryption or Client-Side Field Level Encryption (CSFLE) by redirecting the cryptographic binding logic."
}