Trellis MCP is an interface service that connects AI assistants with Trellis 3D generation models, supporting rapid generation of 3D assets through natural language and importing them into Blender. This project is based on an open - source model and requires self - deployment of the API backend. It is fast and free, but there are stability risks.

This project demonstrates the security risks of using AI agent tools through extendable-agents, especially the potential leakage of sensitive information caused by tool sharing under the MCP protocol, and provides security usage suggestions.

Rug-Check-MCP is an MCP server project used to detect potential risks of Solana meme tokens, helping AI agents avoid rug pull scams and unsafe projects. It retrieves token data through the Solsniffer API and provides structured analysis results, including the name, symbol, risk score, market capitalization, price, supply, risk details, and audit status.

Lilith Shell is an enhanced MCP server that allows AI assistants to execute commands in the terminal, providing security controls and testing capabilities. It should be used in a virtual machine or a resettable development environment due to security risks.

The behavior prediction MCP server provides AI-driven tools for analyzing wallet behavior prediction, fraud detection, and rug pull scam prediction, helping to protect DeFi users and monitor liquidity risks.