Junglewise Threat Intelligence

CVE-2026-61461: langgenius Dify SQL injection in MyScale vector store backend

CVE-2026-61461 · Severity: high · CVSS 8.8 · Published 2026-07-10

Technologies: Langgenius Dify. Vendors: Langgenius.

Executive brief

Dify is an open-source platform used to build and manage AI applications. A security flaw in its MyScale database component allows an attacker with basic user access to run unauthorized database commands. This could lead to the theft of sensitive AI training data, modification of application records, or a complete shutdown of the database service.

Technical details

A SQL injection vulnerability exists in the MyScale (ClickHouse) vector store backend of Dify. The root cause is the direct interpolation of user-controlled search queries into ClickHouse SQL statements using f-strings within the `search_by_full_text` method, without proper escaping or parameterization. An authenticated attacker can exploit this by sending a crafted POST request to the `/v1/datasets/{id}/hit-testing` endpoint with a malicious query string. Successful exploitation allows the attacker to read, modify, or delete data within the underlying ClickHouse database. The issue is fixed in version 1.16.0-rc1.

Affected products

  • langgenius Dify < 1.16.0-rc1

Timeline

  • 2026-07-01: disclosed: Issue reported via GitHub issues
  • 2026-07-02: patched: Fix merged into main branch
  • 2026-07-09: advisory: Release v1.16.0-rc1 published
  • 2026-07-10: disclosed: NVD publication date

References

Related threats