Impact
The vulnerability lies in the MCP server’s handling of the table_name parameter, which is directly interpolated into KQL queries via f‑strings without sanitization. This flaw allows an attacker or a maliciously prompted AI assistant to inject arbitrary KQL code, granting the ability to read, modify, or delete data in the Azure Data Explorer cluster. The weakness is a classic command injection, specifically CWE‑943, and can have serious confidentiality and integrity consequences.
Affected Systems
The affected product is the adx‑mcp‑server maintained by pab1it0. Versions up to and including 0.1.1 are vulnerable; any installation using those releases should be considered at risk.
Risk and Exploitability
With a CVSS score of 8.3 the vulnerability is classified as high severity. The EPSS score is not available, and the flaw has not yet been listed in the CISA KEV catalog. The likely attack vector is through legitimate MCP client traffic or an AI assistant that is given control of the client; the attacker need not have prior credentials beyond the ability to send MCP requests.
OpenCVE Enrichment
Github GHSA