agentguard-python-sdk 0.1.0

A production-grade middleware for AI agents to perform on-chain payments and verifiable consent.

AgentGuard SDK

AgentGuard is a production-grade middleware that enables AI agents to perform on-chain payments and generate verifiable consent proofs in compliance with DPDP (Digital Personal Data Protection) standards.

By isolating private keys in a hardened MCP Server (Vault) and using the AgentGuard Backend (Dispatcher) as a secure proxy, the SDK allows developers to add monetization and compliance to their agents with zero blockchain complexity.

Installation

pip install agentguard

Quick Start (3-Line Usage)

import asyncio
from agentguard import AgentGuardClient

async def main():
    # 1. Initialize the client
    async with AgentGuardClient(wallet_address="YOUR_WALLET_ADDRESS") as client:
        
        # 2. Perform a secure payment (includes automatic DPDP consent hashing)
        receipt = await client.pay_and_fetch(
            resource_url="https://api.stock.com/reliance",
            amount_algo=0.05,
            purpose="Financial Analysis"
        )
        print(f"Payment Successful! TX ID: {receipt.tx_id}")

        # 3. Verify compliance audit trail
        proof = await client.verify(receipt.tx_id)
        print(f"On-Chain Verified: {proof.verified}")

if __name__ == "__main__":
    asyncio.run(main())

Architecture

1. SDK: Generates UUID nonces and SHA256 consent hashes.
2. Backend: Proxies requests to the internal vault and maintains audit records.
3. MCP Server (Vault): A hardened, isolated environment that signs transactions using Algorand.
4. Algorand Blockchain: Provides the irrefutable proof-of-consent and payment settlement.

Features

• Async First: Built on httpx for high-performance agentic loops.
• Zero-Trust: Private keys never leave the vault.
• DPDP Compliant: Automatic cryptographic logging of purpose-bound consent.
• Simple Audit: Direct links to blockchain explorers for every transaction.

---

© 2026 AgentGuard Team. Built for the Algorand Ecosystem.