# Jesse MCP - Phase 2 Implementation Roadmap
**Status**: Ready to begin
**Location**: `~/source/jesse-mcp/`
**Git History**: 3 commits with full history
## Phase 2: Real Jesse Integration
### Objectives
1. Integrate with Jesse's research module
2. Implement real backtest functionality
3. Add strategy management (read/write/validate)
4. Add candle data operations
### Implementation Order
#### 2.1 Jesse Integration Layer (Day 1)
- [ ] Create `jesse_integration.py` module
- Import Jesse research module safely
- Wrapper functions for common operations
- Error handling and validation
**Key files to create:**
```python
# jesse_integration.py
from jesse import research
from jesse.strategies import Strategy
import jesse.helpers as jh
class JesseWrapper:
"""Wrapper around Jesse for MCP operations"""
def __init__(self):
self.jesse_path = '/srv/containers/jesse' # or /mnt/nfs/server1/containers/jesse
def backtest(self, ...):
"""Run backtest via research module"""
def get_strategy(self, name):
"""Get strategy source code"""
def list_strategies(self):
"""List available strategies"""
```
#### 2.2 Real Backtest Tool (Day 1-2)
- [ ] Replace `handle_backtest` placeholder with real implementation
- [ ] Parse backtest config from arguments
- [ ] Call `research.backtest()`
- [ ] Format metrics response
- [ ] Handle errors gracefully
**Example implementation:**
```python
async def handle_backtest(args: Dict[str, Any]) -> Dict[str, Any]:
try:
jesse_wrapper = JesseWrapper()
config = {
'starting_balance': args.get('starting_balance', 10000),
'fee': args.get('fee', 0.001),
'type': args.get('exchange_type', 'futures'),
'futures_leverage': args.get('leverage', 1),
'exchange': args.get('exchange', 'Binance'),
'warm_up_candles': 240
}
routes = [{
'exchange': args.get('exchange', 'Binance'),
'strategy': args['strategy'],
'symbol': args['symbol'],
'timeframe': args['timeframe']
}]
# Get candles
candles = jesse_wrapper.get_candles(
exchange=config['exchange'],
symbol=args['symbol'],
timeframe=args['timeframe'],
start_date=args['start_date'],
end_date=args['end_date']
)
# Run backtest
result = research.backtest(
config=config,
routes=routes,
candles=candles,
hyperparameters=args.get('hyperparameters'),
generate_equity_curve=args.get('include_equity_curve', False),
generate_trades=args.get('include_trades', False)
)
return result
except Exception as e:
return {"error": f"Backtest failed: {str(e)}"}
```
#### 2.3 Strategy Management Tools (Day 2)
- [ ] Implement `handle_strategy_list`
- [ ] Implement `handle_strategy_read`
- [ ] Implement `handle_strategy_validate`
- [ ] Add strategy file parsing
**Operations:**
- Scan `/srv/containers/jesse/strategies/` directory
- Parse Python class structure
- Extract hyperparameters
- Validate syntax before saving
#### 2.4 Candle Data Management (Day 2-3)
- [ ] Implement `candles_import` tool
- [ ] Implement `candles_available` tool
- [ ] Add date range checking
- [ ] Detect gaps in data
**Operations:**
- List available data ranges per symbol/exchange
- Import new candle data from exchanges via `research.import_candles()`
- Check for gaps or missing data
**Supported Exchanges:**
- Binance
- Bitfinex
- Bybit
- Coinbase
- Gate
- Hyperliquid
- Apex
**Implementation Note:**
Jesse already has `research.import_candles()` which downloads directly from exchange APIs:
```python
from jesse import research
result = research.import_candles(
exchange='Binance',
symbol='BTC-USDT',
start_date='2023-01-01',
show_progressbar=False # Set to False for MCP
)
```
This means the MCP server CAN autonomously download new candle ranges!
#### 2.5 Testing & Validation (Day 3)
- [ ] Update test suite for real operations
- [ ] Add integration tests
- [ ] Test with actual Jesse container
- [ ] Validate all metrics formatting
### Key Dependencies
- Jesse research module (already available)
- Jesse strategies directory structure
- Jesse database connectivity
- NumPy for data handling
- Pandas for data analysis
### Success Criteria
- ✅ Real backtest returns complete metrics
- ✅ Strategy operations work on actual files
- ✅ Candle data operations functional
- ✅ All error cases handled gracefully
- ✅ Test suite passes with real Jesse
- ✅ Git commits clean and documented
### Next Steps After Phase 2
- Phase 3: Optimization tools (optimize, walk_forward)
- Phase 4: Analysis tools (analyze_results, metrics)
- Phase 5: Monte Carlo & Pairs Trading
- Phase 6: Autonomous workflows
- Phase 7-8: Polish & Production deployment
### Notes
- Jesse is already deployed at `/srv/containers/jesse`
- Alternative mount at `/mnt/nfs/server1/containers/jesse`
- Test data available for 2023 in container
- All tools should be stateless and reusable
