Developers

Build with AgentTrace

Give an existing agent an onchain identity, a permission firewall and receipt-verified proofs on Monad testnet, without changing how it decides what to do.

SDK · traceCall

One function wraps an existing agent call: simulate through the firewall, send, verify the receipt, anchor the proof.

SDK reference →

MCP server

Any MCP host (Claude Desktop, Cursor, agent runtimes) gets firewall-bound tools. The model cannot step outside the onchain policy.

MCP config →

HTTP API

Public read API for agents, firewalls, executions, proofs and outcomes under /api/v1. No key needed to read.

Endpoints →

Quickstart

The agent keeps its own key and logic. AgentTrace only bounds and records the call.

import { traceCall } from "agenttrace-monad";

const r = await traceCall({
  network: "monad-testnet",
  signer: process.env.AGENT_KEY, // the firewall's executor
  firewallId: 2,
  target, data,                  // the call your agent wanted to make
});
// off-policy calls throw FIREWALL_REJECTED before anything is sent
// r: { txHash, executionId, proofStatus, proofHash, anchorTxHash, proofUrl }

Contracts · Monad testnet · chain 10143

AgentRegistry
AgentFirewall
AgentProof
DemoProtocol

Want to see the full path first? Run the live demo with your wallet, or read the docs.