Give your AI agent
on-chain superpowers.
Your Zafeguard workspace is a fully MCP-compatible server. Any AI agent that speaks the Model Context Protocol — Claude, GPT-4o, or your own — can discover and execute your workflows as native tools, with a full audit trail on every run.
How it looks in practice
{ "workflow": "eth-rebalance-v2", "inputs": { "chain": "arbitrum", "pct": 20 } }{ "workflow": "payroll-weekly", "inputs": { "amount": 500, "token": "USDC" } }Connect in 3 steps
Enable MCP on your workspace
MCP is off by default — flip the toggle in Workspace Settings → "MCP — AI agent access". A dedicated token is generated and shown once; copy it immediately.
POST https://api.zafeguard.com/v1/mcpPoint your client at the endpoint
MCP-specific tokens (zmcp_…) authenticate via the x-mcp-token header — separate from your SDK API key so you can rotate it independently.
x-mcp-token: zmcp_abc12345…Let the agent take action
Your assistant can inspect workflows, execute components, trigger runs, and build new workflows on the canvas — all autonomously, all through the same MCP endpoint.
workflows.execute · components.execute · canvas.create_componentWhat agents can do with Zafeguard
Natural language trading
"Swap 10% of my ETH to USDC" — the agent calls the workflow, you see the on-chain result.
Autonomous DeFi strategies
Agents monitor market conditions and trigger rebalancing workflows automatically, no manual intervention.
AI-powered treasury ops
Schedule recurring payments, route funds, and approve transactions — all driven by your AI agent.
Cross-chain orchestration
Agents reason across chains, pick the cheapest route, and execute multi-step workflows end-to-end.
User intent execution
Parse what a user wants in natural language and translate it into a signed on-chain transaction instantly.
Automated compliance checks
Agents verify wallet status, check allowlists, and enforce on-chain policy before executing any action.