mcp-xray is an open-source security scanner and dynamic penetration testing tool released in July 2026 by Traceforce, a security-focused startup. This tool is designed to detect vulnerabilities in the Model Context Protocol (MCP) infrastructure, which acts as a bridge between large language model (LLM)-based AI agents and local data sources. Similar to how traditional web security scanners meticulously analyze web servers to identify vulnerabilities, mcp-xray scans for anomalies in real-time, including verifying the integrity of configurations on connected MCP servers, checking TLS certificates, detecting exposed loopback addresses, and identifying potential issues that may arise during tool calling. In conventional agent development environments, LLMs often operate with vulnerabilities, such as potential for privilege abuse, remote code execution (RCE), or server-side request forgery (SSRF) attacks, as they autonomously execute tools. In particular, when multiple MCP servers are connected, if the tool description is ambiguously written, it can lead to a "confused deputy" problem, where the agent unintentionally executes unintended parameters or dangerous commands. mcp-xray differentiates itself from existing static analysis tools by going beyond static file inspection to dynamically generate intelligent vulnerability testing plans using LLMs and directly perform penetration testing, proactively identifying and defending against hidden threats in the agent ecosystem. In the fields of biotechnology research and genomic analysis, scenarios are being actively explored where AI agents directly query sensitive patient data or high-value compound library databases to automate exploration. If researchers run the config-scan command of mcp-xray on a custom MCP server they have built for data retrieval, it can detect whether API secret keys for accessing remote databases are hardcoded in plain text within local configuration files and verify input/output vulnerabilities that could trigger path traversal attacks at a baseline level. Finally, the results of the penetration testing are output in the standard security report format (SARIF), allowing it to be integrated into the laboratory's internal security review process, enabling the rapid implementation of a secure research automation infrastructure.

mcp-xray is an open-source security scanner and dynamic penetration testing tool released in July 2026 by Traceforce, a security-focused startup. This tool is designed to detect vulnerabilities in the Model Context Protocol (MCP) infrastructure, which acts as a bridge between large language model (LLM)-based AI agents and local data sources. Similar to how traditional web security scanners meticulously analyze web servers to identify vulnerabilities, mcp-xray verifies the configuration integrity of connected MCP servers and performs TLS