Shellward, from Jnmetacode, is specialized AI agent and Model Context Protocol (MCP) security middleware. Built with TypeScript and running on Node.js, it functions as a standalone MCP server. This platform provides an 8-layer defense system, acting as an AI application compliance gateway. It is distributed as an npm package, making its command-line tools globally accessible upon installation. Shellward helps organizations manage AI agent risks, ensuring adherence to various cybersecurity regulations and data handling policies.

What does Shellward connect to?

Shellward connects to AI tools and platforms that support the Model Context Protocol (MCP), acting as a standalone MCP server. It supports the stdio transport protocol for communication. Confirmed compatible AI clients include Claude Desktop, Claude Code, Cursor, ChatGPT, LangChain, AutoGPT, OpenAI Agents, and Hermes Agent. This enables it to integrate with a range of AI environments, providing security and compliance capabilities across different agent types and development frameworks.

Core features and tools

Shellward offers several callable tools for security and compliance. It provides shellward-mcp for global MCP server functionality. For scanning and reporting, users can employ shellward scan with options like "--json" or "--html report.html". A local graphical interface is available via npx shellward web --local. Automatic MCP configuration is handled by npx shellward init. Internally, it includes MCP tools such as check_command, check_injection, scan_data, and compliance_check for specific security functions.

Who is it for?

Shellward's primary audience includes developers, security teams, and organizations that utilize AI agents. It is particularly useful for those needing to ensure compliance with regulations such as China's cybersecurity law, PIPL, and Equal Protection 2.0, as well as managing data export and AI labeling requirements. Its capabilities address the security and governance needs of AI projects. Skip this if your AI applications do not require strict compliance with data handling or cybersecurity regulations.

How to get started

No credentials are required for its core operation as security middleware. Configuration can be influenced by environment variables such as SHELLWARD_MODE and SHELLWARD_THRESHOLD, and it supports the stdio transport protocol.

Use cases

• Performing a compliance health check on AI projects to identify risks like data export, hardcoded keys, and personal information exposure, generating reports for auditing.
• Providing runtime protection by intercepting and blocking dangerous commands, prompt injection attacks, and data exfiltration during AI agent execution.
• Integrating into CI/CD pipelines to enforce security policies and fail builds if critical issues are detected.

Shellward serves as an essential security and compliance middleware for AI agents operating on the Model Context Protocol. Its TypeScript and Node.js foundation provides an 8-layer defense system, addressing critical needs for data governance and AI application security. It supports various AI clients and offers tools for scanning, reporting, and runtime protection.