Mind Network is deploying its FHE (Fully Homomorphic Encryption) driven secure intelligent agent ecosystem on BytePlus cloud infrastructure, achieving privacy protection and verifiable cross-platform execution for AI Agents.

Today, AI and agents are becoming the next 'app store', transforming the way we live and work. Tools like ChatGPT, DeepSeek, and their plugins have become a part of our daily lives.

There is no doubt that the era of AI Agents has arrived, and the growth rate is rapid.

However, this rapid expansion also brings three fundamental challenges:

  • How do we trust these agents?

  • Are they reliable enough?

  • Are they flexible enough to be applied in the real world?

As a global cloud platform for enterprise clients, BytePlus provides powerful AI capabilities and a mature service architecture, laying a solid foundation for agent deployment. By combining Mind Network's privacy computing and verifiability features, developers can enhance the credibility and autonomy of AI applications on the BytePlus platform. This integration demonstrates how innovative blockchain-AI solutions can independently land on BytePlus to meet the growing demand for secure and reliable AI.

In response to the challenges of trust and verifiability, Mind Network has launched the FHE-driven MCP protocol.

This protocol is now available as a pluggable tool on BytePlus. The MCP can be integrated into various agent platforms, allowing developers and users to deploy AI Agents with cryptographic verification, on-chain consensus mechanisms, and privacy-first guarantees, building a solid and secure foundation for the growing Agentic ecosystem.

Core Solutions

Mind Network has introduced FHE encryption verification capabilities for AI Agents.

Since each Agent may run different backend models, this solution uses FHE to verify and confirm model declarations, establishing trust.

How it works:

  • Node providers can fork verification templates and deploy them as independent verification nodes to BytePlus with custom logic.

  • Each node encrypts verification results using FHE and submits the ciphertext to the verification center, making decisions through the on-chain FHE encryption consensus algorithm.

  • The FHE verification MCP protocol is responsible for on-chain communication between agents.

This solution is modular and pluggable, adaptable for consumer-grade AI Agents, and open to community contributions, achieving decentralized, trustworthy, and reliable intelligent verification.

User Experience

Once launched on BytePlus, users and consumer-grade intelligent agents can connect to Web3 wallets to initiate blockchain-secured FHE verification.

What is MCP

The MCP (Model Context Protocol) plugin developed by Mind Network is built on a general MCP framework and is designed specifically to bridge AI models with blockchain infrastructure. It allows external systems to securely query AI models, ensuring that requests are encrypted and verified while maintaining privacy. MCP aims to minimize the risk of sensitive user data leakage and provides a tamper-proof channel for AI agent interactions, applicable to DApps and enterprise-level scenarios.

The original intention of Mind Network in developing MCP

Mind Network has developed the MCP to enable AI Agents to assist users in complex interactions through natural language in a blockchain-verifiable manner. Users can interact directly with large language models like Llama through their wallets without manually operating the blockchain interface, safely completing tasks such as claiming FHE rewards.

The MCP design ensures that the user's private keys always remain within the wallet and are not exposed to the AI Agent during message signing or transaction processes. Additionally, MCP employs a multi-node FHE voting mechanism to authenticate and verify AI models and their operating environments, ensuring that data and interactions are genuinely tamper-proof and effectively addressing the challenges of trust and verification in decentralized scenarios.

The MCP plugin primarily serves developers, making it more intuitive to develop blockchain-integrated applications and seamlessly access the FHE-driven Agentic World ecosystem.

Four Major Advantages

  • Seamless automation: Users can complete on-chain operations through natural language commands without cumbersome steps.

  • FHE-backed trust: All model calls are certified through multi-node FHE voting, ensuring the integrity of AI and data.

  • Data privacy: Subsequent features will support users in using FHE encryption for session requests, autonomously protecting sensitive details.

  • Scalable agent integration: MCP provides a foundation for future multi-DApp integrated AI assistants, achieving low-threshold task optimization and reward maximization.

Operation guide for MCP plugin in Agent Workspace

The steps are as follows:

  1. Enter Workspace

    • Log in to the Agent Workspace: https://www.coze.com/space

    • Enter the 'Library' and create a new plugin

  2. Plugin Settings

    • Fill in the plugin name and description

    • Select 'Cloud Plugin - Based on Existing Services' for creation method

    • Fill in the Plugin URL: https://mcp-api.mindnetwork.io.

    • Click 'Create Tool'

  3. Configure Tool

    • Fill in Tool Name and Tool Description

    • Fill in Tool Path: https://mcp-api.mindnetwork.io/ai/deepseek/chat

    • Configure input and output parameters

  4. Test Run

    • Click 'Test Run' in the upper right corner

    • Fill in parameters:

      • wallet (wallet address)

      • signature (obtain by signing 'Sign to confirm your DeepSeek query. No gas or fee required.' at https://app.mycrypto.com/sign-message)

    • Optionally use MyCrypto signature; BytePlus does not endorse third-party tools.

  5. Publish Plugin

    • After the test run passes, click 'Publish'

  6. Create and integrate AI Agent

    • Click 'Project Development' in the sidebar → 'Create' → 'Create AI Agent'

    • After creation, add the newly published MCP plugin

  7. Interactive debugging online

    • 'Preview and Debug' page allows interaction by entering wallet and signature

    • After completion, click 'Publish' to launch the Agent

FHE verification nodes in BytePlus function service development guidelines

  • Development can refer to Mind Network's open-source code repository: https://github.com/mind-network/mind-fhe-zk-voting-paillier-ts

  • After local testing passes, go live according to BytePlus FaaS guidelines: https://www.byteplus.com/en/product/faas

  • You will need to select and register a FHE Validation Hub: https://agent.mindnetwork.xyz/agenticworld

  • Complete the launch and registration, and nodes can participate in the AI Agent verification.