Skip to main content
Glama

šŸ›”ļø CHAINZ Sentinel MCP: Binance Agent OS Integration

Binance Agent OS Model Context Protocol NVIDIA NIM License

Binance Agent OS Mini Hackathon Entry
Primary Track: Track B — Connect your MCPs and trade ($40,000 USDC)
Secondary Track: Track A — Build an AI agent with Agent OS ($20,000 USDC)


🌟 Executive Summary

CHAINZ Sentinel MCP is an institutional Model Context Protocol (MCP) server that supercharges Binance Agent OS with sub-millisecond AI trade risk assessment (powered by NVIDIA NIM) and cross-venue zero-debt real-yield arbitrage.

While traditional trading agents execute orders blindly based on raw prompt matching, CHAINZ Sentinel introduces an automated capital preservation guard:

  1. Pre-Trade Risk Sentinel: Before any Binance trade executes, NVIDIA NIM evaluates market volatility and calculates exact liquidation hazard scores in $<50\text{ ms}$.

  2. Natural Language Intent Router: Translates conversational user goals into validated spot and futures trading parameters.

  3. Real-Yield Alpha Arbitrage: Real-time benchmarking comparing Binance Earn / Funding rates with CHAINZ Layer 1 on-chain real-yield vaults (AMM fee sharing + RWA dividends).

  4. Execution Safety Gate: Blocks unsafe trades and automatically downsizes high-leverage orders to prevent liquidation spirals.


Related MCP server: Binance MCP Server

šŸ›ļø System Architecture

flowchart TB
    subgraph Binance_Ecosystem [Binance Agent OS & Client Layer]
        User["User Prompt / Autonomous Trader"] --> AgentOS["Binance Agent OS"]
    end

    subgraph MCP_Layer [Chainz Sentinel MCP Server (JSON-RPC 2.0 stdio)]
        AgentOS <-->|"Model Context Protocol"| MCP["chainz-sentinel-mcp (Node.js)"]
        MCP --> Tool1["evaluate_trade_risk"]
        MCP --> Tool2["parse_trade_intent"]
        MCP --> Tool3["query_yield_arbitrage"]
        MCP --> Tool4["execute_hedged_trade"]
    end

    subgraph Intelligence_And_Execution [AI & Settlement Engines]
        Tool1 & Tool2 <-->|"Inference API"| NIM["NVIDIA NIM GPU Sentinel (integrate.api.nvidia.com)"]
        Tool3 <-->|"Real-Yield Oracle"| L1["CHAINZ Layer 1 Staking Vaults (Chain ID: 999402778)"]
        Tool4 <-->|"Order Routing"| BinanceAPI["Binance Spot / Futures API"]
    end

šŸ› ļø Exposed MCP Tools

1. evaluate_trade_risk

  • Description: Evaluates cryptocurrency trade risk, liquidation probability, and systemic volatility using NVIDIA NIM GPU-accelerated inference.

  • Inputs: symbol (e.g. BTCUSDT), side (BUY/SELL), amount, leverage (default 1), currentPrice (optional).

  • Outputs:

    • risk_score ($0 - 100$)

    • verdict (APPROVED / CAUTION / REJECTED)

    • liquidation_buffer_pct (Distance to liquidation)

    • max_recommended_leverage

2. parse_trade_intent

  • Description: Translates conversational prompts into structured Binance trading parameters.

  • Inputs: prompt (string)

  • Outputs: symbol, action, order_type, amount_usd, leverage, notes.

3. query_yield_arbitrage

  • Description: Benchmarks Binance Earn / Funding rates against CHAINZ zero-debt fee-sharing vaults (55% AMM fees + 30% RWA dividends + 15% bridge relayers).

  • Inputs: symbol (USDT, USDC, BTC, ETH), capitalUsd.

  • Outputs: Baseline yield, CHAINZ yield, Net Alpha APY spread, and recommended capital allocation.

4. execute_hedged_trade

  • Description: Gates order execution against the pre-trade risk score, executing only when capital safety criteria are met.

  • Inputs: symbol, side, amountUsd, riskScore, overrideApproved.

  • Outputs: status (FILLED or REJECTED), order_id, zero_debt_guard status.


šŸš€ Quick Start: Connecting to Binance Agent OS

Prerequisites

  • Node.js v18+ (tested on Node v20/v24)

  • No external npm package installations required! Zero internet drain.

Step 1: Clone or Open the Directory

cd binance-agent-mcp

Step 2: Test the Server Locally

Run the automated MCP test suite to simulate Binance Agent OS:

node test_mcp_client.mjs

Expected output:

======================================================================
šŸš€ Testing Chainz Sentinel MCP Server for Binance Agent OS
======================================================================

1. Testing 'initialize' handshake...
   āœ… Server Name: chainz-sentinel-mcp
   āœ… Protocol Version: 2024-11-05

2. Testing 'tools/list' discovery...
   āœ… Found 4 Tools: evaluate_trade_risk, parse_trade_intent, query_yield_arbitrage, execute_hedged_trade

3. Testing 'evaluate_trade_risk' (NVIDIA NIM GPU Sentinel)...
   āœ… Risk Score: 58 / 100
   āœ… Sentinel Verdict: CAUTION

4. Testing 'parse_trade_intent'...
   āœ… Derived Symbol: ETHUSDT, Action: BUY

5. Testing 'query_yield_arbitrage'...
   āœ… Binance: 7.2% APY vs CHAINZ: 18.4% APY (+11.2% Net Alpha)

6. Testing 'execute_hedged_trade'...
   āœ… Execution Status: FILLED (No liquidation hazard detected)

šŸŽ‰ ALL TESTS PASSED (100% SUCCESS) — READY FOR BINANCE AGENT OS!

Step 3: Add to Binance Agent OS Config

Add the server configuration into your Binance Agent OS MCP registry (binance_agent_config.json):

{
  "mcpServers": {
    "chainz-sentinel": {
      "command": "node",
      "args": ["<PATH_TO_DIR>/index.mjs"],
      "env": {
        "NVIDIA_NIM_BASE_URL": "https://integrate.api.nvidia.com/v1",
        "NVIDIA_API_KEY": "<YOUR_NVIDIA_API_KEY>",
        "NVIDIA_DEFAULT_MODEL": "meta/llama-3.1-8b-instruct"
      }
    }
  }
}

šŸ† Hackathon Alignment & Value Proposition

Evaluation Criteria

How CHAINZ Sentinel Exceeds Expectations

Technical Complexity

Combines the Model Context Protocol (MCP) standard with NVIDIA NIM sub-millisecond GPU inference and EVM Layer 1 smart contracts.

Real Trading Utility

Prevents retail and algorithmic traders from catastrophic liquidation cascades through pre-trade AI safety gating.

Zero External Bloat

Fully self-contained Node.js JSON-RPC stdio engine that adheres to strict zero-dependency bandwidth guidelines.

Cross-Ecosystem Synergy

Bridges Binance's world-leading liquidity with CHAINZ asset-backed zero-debt financial architecture.


šŸ“„ License

MIT License — Free and open-source for the Binance & Web3 AI community.

Available Tools

4 tools
evaluate_trade_riskB

Evaluates cryptocurrency trade risk and liquidation probability using NVIDIA NIM sub-millisecond GPU inference.

Inputs: side (required), amount (required), symbol (required), leverage, +1 more.

execute_hedged_tradeB

Routes a trade with pre-trade risk sentinel gating to prevent liquidation cascades and preserve capital.

Inputs: side (required), symbol (required), amountUsd (required), riskScore (required), +1 more.

parse_trade_intentC

Translates natural language user trading prompts into structured parameters for Binance Agent OS.

Inputs: prompt (required).

query_yield_arbitrageA

Compares Binance Earn and funding rates with CHAINZ on-chain zero-debt real-yield vaults (AMM fee sharing + RWA dividends).

Inputs: symbol, capitalUsd.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.5/5.0

    Scored across 4 tools

    Disambiguation4/5

    Each tool has a generally distinct role: parsing intent, evaluating risk, querying yield opportunities, and executing hedged trades. The only mild overlap is that execute_hedged_trade includes pre-trade risk gating, which could make an agent unsure whether to call evaluate_trade_risk separately.

    Naming Consistency5/5

    All four tools follow the same verb_noun snake_case pattern: evaluate_trade_risk, parse_trade_intent, query_yield_arbitrage, execute_hedged_trade. This makes the tool set predictable and easy to navigate.

    Tool Count5/5

    Four tools is a well-scoped size for a focused risk-sentinel and trade-orchestration server. Each tool provides a distinct, necessary capability without unnecessary bloat.

    Completeness3/5

    The set covers an end-to-end flow from intent parsing to risk evaluation to hedged execution and yield comparison. However, it lacks supporting operations such as trade status lookups, position/balance queries, or risk parameter configuration, so an agent cannot fully manage a trade lifecycle natively.

    Maintenance

    ActivityStale
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI agents to perform cryptocurrency trading operations on Binance exchange through 30 comprehensive tools supporting spot trading, futures contracts, options, account management, market data analysis, and risk control features. Provides enterprise-grade security with local encrypted API key storage and supports multiple account types with sandbox environment testing.
      9
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI assistants to interact directly with the Binance cryptocurrency exchange for trading, portfolio management, market analysis, and strategy automation via natural language.
      142 npm
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables risk-gated AI trading on Binance by researching market data, enforcing policy checks, and building executable MCP order requests that require user confirmation before the official Binance MCP server executes them.
      MIT