- A skill for the OpenClaw AI assistant
- An MCP server to make the Zyfai SDK accessible to AI tools like Claude Code
Why This Matters
Today, most DeFi interactions require a human in the loop: clicking, signing, monitoring. The next phase of onchain finance is autonomous, where AI Agents operate as independent economic actors. Agents will transact with each other, and our agentic yield stack needs to be ready: accessible not only to humans, but to Agents too. For that to happen, yield infrastructure needs to be machine-readable and machine-accessible. This is the foundation of Agent-to-Agent commerce: agent communicating with each other, agents discovering yield, agents executing transactions, agents verifying results via ERC-8004, and agents settling payments via x402. We made this possible by building standardized interfaces:- A SKILL.md for the OpenClaw AI assistant
- An MCP server
OpenClaw Skill
The Zyfai Skill lets any OpenClaw Agent interact with Zyfai’s yield layer through natural language. Once installed, an Agent can query yield opportunities, check portfolio positions, track earnings, and monitor performance across all supported chains. In other words, it is just another way to interact with Zyfai. Instead of using the zyf.ai front-end, you chat with your personal OpenClaw assistant.MCP Server
The Zyfai MCP Server is a production-ready Model Context Protocol server that allows you to leverage Zyfai SDK via chat-bases interactions with AI tools like Claude Code. Available tools include:- Portfolio Management: multi-chain portfolio data, active DeFi positions
- Opportunities Discovery: Conservative and Aggressive yield strategies, available protocols and pools per chain
- Analytics & Metrics: onchain earnings, daily earnings, TVL, volume, active wallets, rebalance frequency
- Historical Data: transaction history, daily APY tracking, APY per strategy