Horus Flow MCP

What horustechltd/horus-flow-mcp MCP server does

horustechltd/horus-flow-mcp MCP server delivers live order-flow analysis for cryptocurrency and US equity markets. Its stated inputs include Binance L1/L2 orderbook data and Alpaca SIP equity streams. The system is designed to turn raw market activity into structured signals that an AI agent can use as context.

The analysis focuses on market microstructure rather than indicators such as RSI or MACD. Reported outputs can include a signal, confidence value, market state, risk level, timestamp, bid/ask ratio, buy/sell ratio, five-second delta, whale-activity status, wall side, and event flags. The README also shows outputs for liquidity events, aggressive buying or selling, and possible spoofing conditions.

How it works

The architecture describes separate ingestion paths for Binance and Alpaca WebSocket feeds. An imbalance calculator evaluates bid and ask conditions, while a flow calculator tracks tape deltas. These results pass through rules for behavioral patterns such as spoofing and liquidity changes.

An in-memory prediction tracker is included in the described processing flow. It feeds results back into the rules layer for accuracy validation. The processed data is then delivered to an MCP context, a FastAPI client, and a real-time dashboard according to the architecture shown in the README.

The example output is JSON-shaped and includes both a high-level decision and supporting metrics. The README reports a 29 ms latency badge and describes local response times as sub-millisecond in an audit example, but these figures should be treated as project-reported results rather than guaranteed performance for every deployment or market condition.

Setup and configuration

The documented quickstart requires Python 3.12 or newer and installs the mcp and httpx dependencies:

pip install mcp httpx

Live data access requires the RAPIDAPI_KEY environment variable. The README points users to the Horus Flow Intelligence API on RapidAPI for obtaining a key. It also references a developer portal and a Postman collection for direct API testing.

The example starts the server with SSE transport on port 8011:

python horus_mcp_public.py --transport sse --port 8011

The supplied excerpt does not document a complete MCP client configuration, repository installation package, or additional environment variables.

Tools and capabilities

The documented capabilities include:

• Inspecting Binance cryptocurrency orderbook data.
• Reading Alpaca SIP US equity market data.
• Calculating bid/ask imbalance and order-flow ratios.
• Tracking short-term flow deltas, including a five-second delta.
• Identifying liquidity events and liquidity withdrawal patterns.
• Applying rules for spoofing indicators, wall placement, and aggressive taker activity.
• Returning structured signals with confidence, risk, market-state, and supporting metrics.
• Providing processed results through MCP, a FastAPI client, and a real-time dashboard.

The material does not provide a formal list of MCP tool names or their exact input schemas, so integrations should be tested against the repository before relying on specific tool names.

Limitations and notes

horustechltd/horus-flow-mcp MCP server requires access to live market data through a RapidAPI key. The README describes a $29 Pro plan, but the available material does not establish whether a free tier exists or whether all usage requires payment. It also does not document trading execution, broker order placement, or guaranteed prediction accuracy; the described role is analysis and signal generation.

The README includes an audit example using BTCUSDT and mentions symbol handling for SOLUSDT, but coverage and availability for other symbols are not specified. Its claims about institutional behavior, confidence, latency, and accuracy are project-reported examples rather than independently verified guarantees.