SQL Injection and Security Boundary Bypass in Google BigQuery Forecasting Tool
CVE-2026-15829

8.6HIGH

What is CVE-2026-15829?

A SQL injection and security boundary bypass vulnerability exists within the prebuilt BigQuery forecasting tool from Google. This issue arises from the tool's handling of client-controlled parameters such as data_col, timestamp_col, and id_cols, which are processed as plain strings and incorporated without proper escaping into a generated SELECT statement. Although an allowedDatasets mechanism is employed to restrict queries, it only validates the history_data parameter; the assembled query is ultimately executed without re-validation. This flaw can be exploited by an attacker to bypass the configured allowedDatasets boundary, potentially allowing unauthorized access to read arbitrary BigQuery tables through crafted multi-statement or cross-dataset query blocks.

Affected Version(s)

MCP Toolbox for Databases (googleapis/mcp-toolbox) 0.13.0 <= 1.3.0

References

CVSS V4

Score:
8.6
Severity:
HIGH
Confidentiality:
High
Integrity:
High
Availability:
None
Attack Vector:
Network
Attack Complexity:
Low
Attack Required:
None
Privileges Required:
Undefined
User Interaction:
None

Timeline

  • Vulnerability published

  • Vulnerability Reserved

Credit

Matteo Panzeri (UniversitĂ  di Pavia), GitHub @matte1782
.