The vulnerability resides in the scan_vulnerabilities component of Vorota-ai nmap-mcp, where the allowlist for NSE script categories is improperly validated. Attackers who can influence the MCP client or inject prompts can supply categories such as exploit, dos, brute or intrusive. These categories enable the execution of potentially malicious NSE scripts against scan targets, potentially leading to exploitation or denial‑of‑service of those targets.

All versions of the nmap-mcp tool from Vorota-ai up to and including commit 3636afa are affected. No specific version numbers are provided beyond the commit hash.

With a CVSS score of 4.8, the vulnerability poses a moderate risk. The EPSS score is not available, and it is not listed in CISA’s KEV catalog. Based on the description, the attack vector likely requires an attacker to control or influence the MCP client or inject prompts; an attacker with such access can directly trigger the execution of blocked script categories. The impact is limited to the systems scanned by the tool but the ability to run exploit and intrusion scripts can lead to compromise of those targets.