🦅 Horus Flow MCP Server

👁 Smithery Badge
👁 Glama Quality
👁 Security: No Vulnerabilities
👁 Python 3.12+
👁 License: MIT

🧠 Why Horus? The AI Agent's "Nervous System"

In the world of autonomous trading, an AI Agent without real-time orderflow is like a pilot flying blind. Technical indicators (RSI, MACD) are lagging echoes of the past. Horus is the Nervous System that provides:

• Proprioception: Real-time awareness of market "muscle" (Orderbook Depth).
• Reflexes: Sub-second detection of institutional "pain" (Liquidity Collapses).
• Intelligence: A context-aware decision matrix that translates raw physics into actionable signals.

"If you want your AI Agent to trade like a whale, you must give it the whale's eyes. Horus is that vision."

🧪 Manus AI Audit: Live Performance Proof

Horus Flow MCP has been rigorously audited by Manus AI (an autonomous execution agent). The results confirm that Horus isn't just code—it's a high-performance reality.

Audit Highlights (Live Test on BTCUSDT):

• Signal Accuracy: Successfully detected STRONG_SELL_PRESSURE with a 0.85 Confidence Score.
• Physics Engine Validation: Confirmed bid_ask_ratio of 0.156, proving the engine's ability to read thin limit books and aggressive selling.
• Zero-Hallucination Logic: The MCP server correctly identified symbols in "Data Gathering" mode (e.g., SOLUSDT), preventing the AI from making false assumptions.

💰 The $29 Arbitrage: Institutional Intelligence for the Price of a Lunch

Why pay thousands for institutional terminals when you can get the same Microstructure Intelligence for a fraction of the cost?

🛑 Stop Predicting Candles. Start Measuring Gravity.

Retail traders use trailing indicators to guess what the market will do based on the past. Horus uses Level 2 Orderbook physics, Tick imbalances, and 5-Second Flow Deltas to measure exactly what institutional whales are doing right now.

Horus doesn't ask "Are we overbought?". It measures gravitational pull and tells you: "Whales are spoofing the bid and aggressive takers are tearing through the ask. Liquidity is collapsing. Bailout now."

⚡ Why AI Agents (Claude, Cursor) Love Horus

If you hook up an AI Agent to a basic technical indicator feed, it gets confused by noise. If you feed an AI Agent Horus, it gets an institutional grade decision matrix.

Look at what Horus catches in real-time during a Liquidity Withdrawal / Flash Crash attempt:

{
 "symbol": "BTCUSDT",
 "signal": "LIQUIDITY_EVENT",
 "confidence": 0.99,
 "market_state": "DISTRIBUTION",
 "risk": "EXTREME",
 "description": "Liquidity withdrawn under price. Aggressive taking.",
 "metrics": {
 "bid_ask_ratio": 7.934,
 "buy_sell_ratio": 0.393,
 "delta_5s": -83040.05,
 "whale_activity": true,
 "large_sell_count": 4,
 "delta_accel": 3.8,
 "wall_side": "BID",
 "wiseman_climate": {
 "market_mode": "CHOP",
 "health": "FRAGILE",
 "confidence": 0.85
 },
 "flags": [
 "GLOBAL_LIQUIDITY_EVENT",
 "SPOOFING_DETECTED(wall=BID)"
 ]
 },
 "timestamp": 1776107738.975
}

The Institutional Alpha:

1. bid_ask_ratio: 7.934 & wall_side: BID: Massive spoofed bid walls are placed by whales below the market to create fake support.
2. buy_sell_ratio: 0.393: Horus sees through the spoofing (SPOOFING_DETECTED flag). It measures real taker flow and discovers aggressive selling is devastating the orderbook.
3. delta_accel: 3.8 & whale_activity: true: Selling momentum accelerated by 3.8x natively tracking 4 major whale dumps within milliseconds.
4. wiseman_climate: FRAGILE: Integrates perfectly with the overarching Horus SaaS macro brain, verifying that Bitcoin's holistic environment is fragile before attacking.
5. The Verdict: With a 0.99 Confidence factor, the AI engine triggers LIQUIDITY_EVENT. It front-runs the ensuing 1-minute crash. This is Institutional Grade.

🏗️ Architecture

graph TD
 %% Styling
 classDef crypto_stream fill:#F3BA2F,stroke:#000,color:#000,stroke-width:2px;
 classDef equity_stream fill:#000,stroke:#09b533,color:#09b533,stroke-width:2px;
 classDef compute fill:#1A1F36,stroke:#00D6FF,color:#fff,stroke-width:2px;
 classDef mcp fill:#632CA6,stroke:#fff,color:#fff,stroke-width:2px;
 classDef client fill:#FF3366,stroke:#fff,color:#fff,stroke-width:2px;
 %% Ingestion
 subgraph Data_Pipelines [Sub-Millisecond Websocket Ingestion]
 B[Binance WSS <br/> L1/L2 Book]:::crypto_stream
 A[Alpaca WSS <br/> SIP Equities]:::equity_stream
 end
 %% Engine
 subgraph Core_Engine [The Physics Engine]
 IC[Imbalance Calculator <br/> Bid/Ask Spread]:::compute
 FC[Flow Calculator <br/> Tape Deltas]:::compute
 Tkr[Prediction Tracker <br/> In-Memory Evaluator]:::compute
 BC[Behavioral Court <br/> Spoofing & Liquidity Rules]:::compute
 end
 %% Output
 subgraph Output_Layer [Data Shield & Delivery]
 MCP[AI Agent MCP Context <br/> Context-Aware Prompts]:::mcp
 API[FastAPI Client <br/> Safe Shielded Outputs]:::client
 Dash[Real-time Dashboard <br/> Live Edge Proof]:::client
 end
 B --> IC
 A --> IC
 B --> FC
 A --> FC
 IC --> BC
 FC --> BC
 BC --> Tkr
 Tkr -- "Validates 1M accuracy" --> BC
 BC --> MCP
 BC --> API
 BC --> Dash

🚀 Quickstart for Trading Bots

Getting started with Horus takes less than 60 seconds. ⚡ Test Instantly: Download our Postman Collection to ping the API directly from your browser.

1. Fire up the Core Engine

# Install Dependencies
pip install mcp httpx
# Set your RapidAPI Key (Required for the server to fetch live data)
export RAPIDAPI_KEY="your_actual_rapidapi_key"
# Run the SSE Server
python horus_mcp_public.py --transport sse --port 8011
import requests
# Connecting to your local Horus instance
response = requests.get("[http://127.0.0.1:8011/v1/flow/crypto/BTCUSDT](http://127.0.0.1:8011/v1/flow/crypto/BTCUSDT)")
flow_data = response.json()
# The Zero-Human Decision Loop:
if flow_data.get('signal') in ['EMERGENCY_DUMP', 'BAILOUT', 'LIQUIDITY_EVENT']:
 print(f"⚠️ {flow_data['symbol']} Market gravity collapsing. Executing Bailout.")
 # execute_market_sell()

🌐 Live Real-Time Dashboard

Horus comes with a stunning glass-morphism WebSocket dashboard out-of-the-box (/dash/). It features a Local Edge Proof Tracker that strictly measures 1-Minute Candle directional predictions generated by the Physics Engine with an openly transparent win rate.

Built with absolute precision for those who need to see the Matrix.
— HORUS INTELLIGENCE 👁️ mcp-name: io.github.horustechltd/horus-flow-mcp