Skip to content

AI Treasury Agent for MNEE Stablecoin - Telegram & Slack bots with gasless payments

License

Notifications You must be signed in to change notification settings

Amethyst001/MNEE-Sentinel

Repository files navigation

MNEE Sentinel

MNEE Sentinel solves the $2T B2B trust gap by moving from human-initiated payments to verifiable agentic mandates. It is the first 'Treasury-in-a-Box' that uses ZK-proofs to ensure payment only follows performance, gaslessly.


MNEE Sentinel is an autonomous B2B Agentic Treasury and Escrow system. It allows businesses (via Telegram and Slack) to authorize an AI Agent to negotiate, verify, and settle payments for digital services using the MNEE Stablecoin without manual gas management or trust-gap risks.

🏆 Key Features

1. Enterprise Audit Logger

Every transaction generates a signed JSON-LD metadata file containing the LLM prompt, ZK-proofs, and the on-chain TxID. This ensures full corporate compliance and a verifiable audit trail.

2. Intent-Based Mandates

Leveraging AP2 (Agent Payment Protocol), the Agent cannot move money without a cryptographically signed "Intent Mandate" that includes a strict Time-to-Live (TTL). Security is baked into the protocol, not just the code.

🏆 The 9 Pillars of Complexity

We didn't just build a bot. We built a Financial Operating System for the Agentic Economy.

Pillar Tech Description
1. Adversarial Governance Gemini Auditor Red Team vs Blue Team AI. Forensic Policy RAG & Velocity Checks.
2. ZK Invoice Privacy Groth16 Simulated Zero-Knowledge Proof structure for privacy preservation.
3. Agent Negotiation Node.js API Autonomous HTTP negotiation with Supplier Agents to cut costs.
4. MNEE Paymaster ERC-4337 100% Gasless User Experience (Relayer architecture).
5. AP2 Mandates EIP-712 Cryptographically secured payment intents with strict TTL.
6. Enterprise Audit JSON-LD W3C-compliant audit trails for corporate compliance.
7. Multi-Sig Reactions Telegram/Slack Emoji-based consensus for high-value transaction approval.
8. Gas Optimization Real-Time Data Monitors Base L2 gas prices to suggest optimal payout times.
9. On-Chain Reputation Credit Logic Dynamic "Trust Score" for agents based on audit pass rates.

🏗 Architecture

  • Interface: Telegram/Slack Bots - The Command Center.
  • Agent Logic: Gemini 2.5 Flash + AP2 - The Brains.
  • Blockchain: MNEE Stablecoin + Safe Core SDK - The Rails.
  • Trust: ZK-Escrow (Simulated) - The Guarantee.

📂 Repository Structure

  • /contracts: Solidity Smart Contracts (ZK-Escrow, MandateRegistry).
  • /agent: Agentic logic, AP2 mandates, TLSNotary verification.
  • /bot: Telegram command center (User Session management).
  • /docs: Executive Audit Trail (Transaction receipts).

🚀 Quick Start

Prerequisites

  • Node.js 18+
  • Telegram account
  • Azure account (for Voice Biometrics) - optional
  • Gemini API key (from aistudio.google.com)

Setup

# 1. Install dependencies
cd agent && npm install && cd ..
cd bot && npm install && cd ..

# 2. Configure environment
cp .env.example .env
# Edit .env with your API keys

# 3. Run the bots (Concurrent)
cd bot && npm run start

Test Commands (in Telegram)

  • /start - Initialize the bot
  • /pay 40 MNEE to @vendor for server costs - Test payment flow
  • Send a voice message - Test biometric authentication

⚠️ Important: Custodial Prototype

This is a hackathon prototype demonstrating a custodial treasury model:

  • The operator holds the treasury wallet private key
  • Users authorize payments through the bot; the operator's wallet executes them
  • For production, consider non-custodial approaches (WalletConnect, Account Abstraction with user-owned keys)

This architecture is suitable for:

  • B2B treasury management where a company controls the wallet
  • Demonstration and proof-of-concept purposes
  • Internal corporate expense automation

📄 License

MIT

🔗 Links

About

AI Treasury Agent for MNEE Stablecoin - Telegram & Slack bots with gasless payments

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published