What is Agent Fox?
Agent Fox is an autonomous trading companion that monitors 100+ Solana tokens around the clock. It calculates technical indicators, generates actionable buy/sell signals, and tracks every prediction against real market outcomes — wins and losses, fully transparent.
No cherry-picked results. No black-box strategies. Just data.
Live at agentfox.trade
Features
Signal Generation
Generates LONG signals with precise entry, three take-profit targets, and stop-loss levels. Each signal includes a confidence score based on indicator confluence and historical accuracy.
RSI Scanner
Scans top 100 Solana tokens across 1H, 4H, and 1D timeframes. Surfaces oversold bounces and overbought reversals with multi-timeframe alignment scoring.
Volume Scanner
Detects volume spikes (2x+ above baseline) in real time. Classifies each spike as bullish, bearish, or neutral with severity ranking and optional Telegram alerts.
Backtesting Engine
Eight trading strategies tested against historical OHLCV data. Reports win rate, profit factor, Sharpe ratio, and max drawdown for each.
Performance Tracking
Every signal is scored against real market outcomes at 24h, 48h, and 7d intervals. Dashboard displays win rate, PnL, and profit factor — updated continuously.
Autonomous Agent (ElizaOS)
Built on ElizaOS v1.7.2. Engages with the Colosseum hackathon forum, posts market updates, and reasons about trades using Claude AI.
Architecture
┌─────────────────────────────────────────────────────────┐
│ Website (Next.js 15) │
│ www.agentfox.trade · Vercel │
└──────────────────────┬──────────────────────────────────┘
│ /api proxy
┌──────────────────────▼──────────────────────────────────┐
│ API Server (Hono) │
│ Railway · Cron Scheduler │
├─────────┬──────────┬──────────┬──────────┬──────────────┤
│ Signals │ RSI Scan │ Volume │ Backtest │ Performance │
│ Engine │ (100 tkn)│ Scanner │ Engine │ Tracker │
└────┬────┴────┬─────┴────┬────┴──────────┴──────────────┘
│ │ │
┌────▼─────────▼──────────▼──────────────────────────────┐
│ Data Layer (Multi-Source) │
│ Jupiter · DexScreener · GeckoTerminal · Birdeye │
└─────────────────────────────────────────────────────────┘
Tech Stack
| Layer | Technology |
|---|
| API | TypeScript, Hono, node-cron |
| Frontend | Next.js 15, React 19, Tailwind, Shadcn UI |
| Agent | ElizaOS v1.7.2, Anthropic Claude |
| Data | Jupiter, DexScreener, GeckoTerminal, Birdeye |
| Indicators | RSI, MACD (3,10,16), EMA, Support/Resistance, Fibonacci, Volume |
| Storage | better-sqlite3, JSON persistence |
| Deploy | Railway (API + Agent), Vercel (Website), Docker |
API
Base URL: https://trading-caller-production-d7d3.up.railway.app
Signals
GET /signals/latest Latest trading signals
GET /signals/performance Win rate, PnL, profit factor
GET /signals/tracked Currently tracked signals
RSI Scanner
GET /rsi/multi All tokens, all timeframes
GET /rsi/oversold Tokens with RSI ≤ 20
GET /rsi/overbought Tokens with RSI ≥ 80
GET /rsi/:symbol Single token across 1H/4H/1D
Volume
GET /volume/signals Volume-based trading signals
GET /volume/spikes Recent spike history
GET /volume/top Top 10 volume movers
Backtesting
GET /backtest/strategies Available strategies
GET /backtest/results Historical results
POST /backtest/run Run a backtest
Market Data
GET /funding/:symbol Funding rates + squeeze alerts
GET /gecko/trending Trending Solana pools
GET /leaderboard/tokens Token performance ranking
| Method | Endpoint | Description |
|---|
GET | /signals/latest | Latest signals with optional action filter |
GET | /signals/history | Paginated signal history |
GET | /signals/performance | Performance metrics |
GET | /signals/tracked | Active tracked signals |
POST | /signals/reset | Reset signal history |
GET | /tokens/:symbol/analysis | Deep single-token analysis |
GET | /rsi/multi | Multi-timeframe RSI for all tokens |
GET | /rsi/scan | Latest scan results |
GET | /rsi/oversold | Oversold opportunities |
GET | /rsi/overbought | Overbought setups |
GET | /rsi/:symbol | Single token RSI |
POST | /rsi/multi/scan | Force rescan |
GET | /volume/status | Scanner status |
GET | /volume/top | Top volume movers |
GET | /volume/spikes | Spike history |
GET | /volume/signals | Volume signals |
POST | /volume/scan | Trigger manual scan |
GET | /funding/:symbol | Funding rate analysis |
GET | /funding/alerts/squeeze | Squeeze alert list |
Quick Start
# Clone
git clone https://github.com/breakthesimulation/trading-caller.git
cd trading-caller
# Install
npm install
# Configure
cp .env.example .env
# Add your API keys to .env
# Run API server (includes scheduler)
npm start
# Run website locally
cd website && npm install && npm run dev
Environment Variables
ANTHROPIC_API_KEY= # Claude AI (signal reasoning)
TELEGRAM_BOT_TOKEN= # Telegram alerts (optional)
TELEGRAM_CHAT_ID= # Telegram target chat (optional)
PORT=3000 # API server port
Project Structure
├── api/ API server (Hono routes, scheduler, WebSocket)
├── research-engine/ Signal generation, technical indicators, data providers
├── volume-scanner/ Real-time volume spike detection
├── oversold/ RSI-based oversold/overbought scanner
├── backtesting/ Historical strategy validation engine
├── performance/ Signal outcome tracking and scoring
├── learning/ Pattern analysis and weight adjustment
├── agent/ Colosseum hackathon agent (heartbeat, forum, brain)
├── eliza/ ElizaOS autonomous trading agent
├── website/ Next.js frontend (Vercel)
├── db/ Data persistence layer
├── scripts/ Utility and registration scripts
└── data/ Runtime data storage
How Signals Work
- Scan — Continuously monitors 100+ Solana tokens for technical setups
- Analyze — Calculates RSI, MACD, EMA, support/resistance, volume, and funding rates
- Weight — Applies learned weights from historical signal accuracy
- Score — Rates confidence based on indicator confluence
- Signal — Publishes actionable call with entry, three targets, and stop-loss
- Track — Monitors outcomes at 24h, 48h, and 7d and updates performance metrics
Links
