Scan Your MCP Security with Proximity’s Open-Source Tool
▼ Summary
– Proximity is an open-source tool that scans Model Context Protocol (MCP) servers to identify exposed prompts, tools, and resources and assess their security risks.
– The tool was created to address the increased attack surface from the rapid adoption of MCP, which can expose clients to risks like prompt injection, tool poisoning, or data exfiltration.
– It works by probing remote or local MCP servers to list exposed elements and can be paired with NOVA, a rule engine that detects suspicious content using pattern-based rules.
– Attackers can weaponize exposed tool descriptions to understand and manipulate systems, making Proximity and NOVA essential for detecting harmful content before deployment.
– Proximity is part of a broader effort to help security teams evaluate AI systems and is available for free on GitHub.
Scanning the security of your Model Context Protocol (MCP) servers is now more accessible thanks to Proximity, a newly released open-source tool. This utility probes MCP servers, whether remote or local, to catalog the prompts, tools, and resources they expose. By identifying these elements, Proximity helps assess potential security vulnerabilities before deployment. It integrates seamlessly with NOVA, a rule engine designed to detect threats such as prompt injection, tool poisoning, and data exfiltration.
The rapid adoption of MCP over the past year has enabled developers worldwide to build and share their own servers, significantly expanding AI functionality. However, this growth has also widened the attack surface, leaving client systems exposed to various risks. Thomas Roccia, the creator of Proximity, emphasized that the tool was developed to offer an initial security review of MCP servers, helping organizations mitigate dangers before integrating them into operational environments.
Proximity operates by systematically querying MCP servers and generating a detailed inventory of accessible components. Each resource typically includes a tool name and description, information that malicious actors could weaponize to understand and exploit system behavior. When used alongside NOVA, the tool enables security analysts to define custom, pattern-based rules. These rules help flag suspicious content, providing an additional layer of defense by identifying possible jailbreak attempts or other harmful activities early in the evaluation process.
Roccia further explained that attackers often leverage exposed tool descriptions to map out system operations and identify weak points. Proximity, in combination with NOVA, allows teams to scan these descriptions proactively, detecting risky configurations or content before they are introduced into a live setting. This approach forms part of a broader initiative to equip security professionals with practical, evolving toolkits that keep pace with the dynamic nature of AI ecosystems.
Proximity is freely available on GitHub, providing an accessible resource for security teams focused on safeguarding AI-augmented workflows.
For ongoing updates on essential open-source cybersecurity tools, consider subscribing to specialized monthly newsletters that focus on the latest developments and resources in the field.
(Source: HelpNet Security)
