CVE-2026-15829
HighCVSS 8.6Exploitation Probability (EPSS)
Low risk7th percentile - higher than 7% of all known CVEs
Summary
SQL injection and security boundary bypass vulnerability in the bigquery-forecast tool of googleapis/mcp-toolbox. An attacker can inject malicious SQL queries via client-controlled parameters, bypassing the allowedDatasets mechanism and accessing arbitrary BigQuery tables.
Risk Assessment
Unauthorized access to sensitive data in BigQuery, potential data leakage or modification.
Recommendation
Update googleapis/mcp-toolbox to a patched version or implement input validation and sanitization before using parameters in SQL queries.
Other vulnerabilities in googleapis/mcp-toolbox
See all- CVE-2026-11720Critical
A path traversal vulnerability exists in the HTTP tool URL builder of googleapis/mcp-toolbox. By supplying path parameters with directory traversal sequences (../), an attacker can bypass configured path restrictions and access unauthorized endpoints on the same target host.
- CVE-2026-11718Critical
An authentication bypass vulnerability exists in the opaque token validation path of googleapis/mcp-toolbox. This flaw allows the acceptance of tokens issued by unauthorized identity providers when the external OAuth provider's response omits the 'iss' field.
- CVE-2026-11717Critical
An authentication bypass vulnerability exists in the googleapis/mcp-toolbox that allows bypassing authentication. The issue is related to opaque token validation, where the absence of the 'active' key in the OAuth 2.0 introspection response results in the acceptance of tokens that should not be considered valid.
- CVE-2026-16481High
The cloud-healthcare-fhir-fetch-page tool in googleapis/mcp-toolbox is vulnerable to SSRF and credential exfiltration. An unvalidated pageURL parameter triggers HTTP requests with an Authorization: Bearer header to arbitrary hosts, leaking OAuth tokens.
Original NVD description (English source)
A SQL injection (CWE-89) and security boundary bypass (CWE-863) vulnerability exists in the prebuilt BigQuery forecasting tool (bigquery-forecast) of googleapis/mcp-toolbox. The tool accepts client-controlled parameters (data_col, timestamp_col, and id_cols) as plain strings and interpolates them unescaped via fmt.Sprintf directly into a generated AI.FORECAST table-valued SELECT statement. While MCP Toolbox utilizes an allowedDatasets mechanism to restrict queries, this defense only validates the history_data parameter; the final assembled query is executed without re-validation. An attacker can break out of the string literal fields (such as timestamp_col) to inject a valid multi-statement or cross-dataset query block. This allows an unauthorized user to bypass the operator-configured allowedDatasets boundary and read arbitrary BigQuery tables.