# AlphaLoop Arc — Autonomous AI FX Prime Broker > Agora Agents Hackathon · Canteen × Circle · Arc Testnet AlphaLoop is a fully autonomous multi-agent prime broker that executes **AI-driven FX trades** on Arc — Circle's stablecoin-native L1. Agents analyze market signals and settle USDC/EURC currency pairs on-chain with sub-second finality. ## Architecture ``` Scout Agent → reads macro signals, RSI, momentum, news Risk Agent → sizes position, approves/rejects trade Learning Agent → adapts strategy based on outcome history Execution Agent → settles USDC ↔ EURC FX trade on Arc ``` External agents (Alice, Bob, Charlie) pay via x402 micropayments and receive fully autonomous FX execution — no humans in the loop. ## What Makes It Agentic - **Scout** pulls live price feeds, RSI, momentum, and Polymarket sentiment - **Risk** runs Kelly Criterion sizing and approves/rejects based on portfolio heat - **Learning** tracks win rates per strategy and adjusts allocation over time - **Execution** settles two on-chain transactions per trade: fee payment + FX delivery Every cycle is fully autonomous. No human triggers any trade. ## Stack | Layer | Tech | |-------|------| | Settlement | Arc Testnet (Chain ID: 5042002) | | FX Pair | USDC ↔ EURC (Circle native stablecoins) | | Broker API | FastAPI | | Agent Interface | MCP Server | | Signal Engine | Python — RSI, VWAP, Bollinger Bands, momentum | | Payment | x402 micropayments | ## API Endpoints | Endpoint | Cost | Description | |----------|------|-------------| | `GET /preview/{asset}` | Free | Price + confidence preview | | `POST /signal` | $0.01 USDC | Full directional signal | | `POST /validate` | $0.02 USDC | Risk Agent validation | | `POST /execute` | $0.05 USDC | FX trade execution | | `POST /broker` | $0.02 USDC | Full 4-agent pipeline | | `GET /status` | Free | Portfolio + performance | | `GET /activity` | Free | Live trade feed | ## FX Settlement Flow ``` Agent pays $0.05 USDC → Broker (x402 fee) [TX 1] Scout scores signal → Risk sizes position Broker settles EURC → Agent (FX delivery) [TX 2] Learning Agent records outcome → adjusts strategy ``` Every trade = 2 confirmed on-chain transactions on Arc. ## Circle Tools Used - **USDC** — native gas token + payment settlement - **EURC** — FX delivery token (EUR-denominated) - **Arc Testnet** — sub-second finality, $0.01 fees ## Live Demo **Broker:** `https://[your-domain]/` **Explorer:** [Broker wallet on ArcScan](https://testnet.arcscan.app/address/0xB9d62aFC43C1d62887AdDBBD159A5F774209e40d) ## Agents | Agent | Asset | Strategy | |-------|-------|----------| | Alice | EURUSD | Balanced — alternates long/short | | Bob | EURUSD | Balanced — 360s cycle | | Charlie | EURUSD | Balanced — 420s cycle | Each agent runs independently, pays its own fees, and accumulates its own FX position history. ## Running Locally ```bash git clone https://github.com/davieslennox0/arc-alpha.git cd arc-alpha pip install -r requirements.txt cp .env.example .env # fill in your keys python3 prime_broker.py # starts broker on port 8002 python3 demo_agents.py # starts Alice, Bob, Charlie ``` ## Hackathon Built for the [Agora Agents Hackathon](https://agora.thecanteenapp.com) by Canteen × Circle · May 2026. **Team:** [@Syke0x](https://x.com/Syke0x)