Jesse MCP Server

# Trading Strategy Development PRD v1.0.0 ## 🎯 Global Architecture & Interfaces ### Shared Data Structures - **Strategy Schema**: Standardized format for strategy definitions - **Performance Metrics**: Common KPIs and measurement standards - **Risk Parameters**: Unified risk management configuration - **Integration Points**: Defined interfaces between agent outputs ### Common Tool Interfaces - **Strategy Development**: `strategy_*`, `optimize`, `walk_forward` - **Backtesting & Analysis**: `backtest`, `backtest_batch`, `analyze_results` - **Risk Management**: `monte_carlo_*`, risk assessment tools - **Market Data**: `get_asset_price`, `list_assets`, news sources ### Inter-Agent Communication Protocols - **Data Transfer**: JSON-based shared state files - **Progress Signaling**: File-based status updates - **Dependency Management**: Clear input/output contracts - **Version Compatibility**: Schema validation and migration --- ## 🤖 Agent A: Strategy Development Section ### Scope - **Primary Responsibility**: Strategy creation, validation, and optimization - **Core Tools**: `strategy_*`, `optimize`, `walk_forward`, `backtest_comprehensive` - **Deliverables**: - Complete strategy implementations - Optimization results and parameter recommendations - Validation reports with performance metrics - Strategy documentation and deployment guides ### Integration Points - **Provides To**: Agent B (strategies for backtesting) - **Receives From**: Agent C (risk constraints and market context) - **Shared Outputs**: Strategy files, optimization parameters, performance baselines ### Success Criteria - **Code Quality**: Clean, documented, error-free implementations - **Performance**: Meets baseline metrics in initial testing - **Validation**: Passes all risk and compliance checks - **Documentation**: Complete with clear usage instructions --- ## 📊 Agent B: Backtesting & Analysis Section ### Scope - **Primary Responsibility**: Comprehensive backtesting, performance analysis, and statistical validation - **Core Tools**: `backtest`, `backtest_batch`, `analyze_results`, `backtest_monte_carlo`, `correlation_matrix` - **Deliverables**: - Detailed backtest reports with key metrics - Performance analysis across market conditions - Statistical significance validation - Comparative analysis between strategies - Risk-adjusted performance metrics ### Integration Points - **Receives From**: Agent A (strategies for testing) - **Provides To**: Agent C (performance data for risk assessment) - **Shared Outputs**: Backtest results, performance metrics, analysis reports ### Success Criteria - **Coverage**: Test across multiple timeframes and market regimes - **Statistical Validity**: Statistically significant results (p < 0.05) - **Risk Analysis**: Comprehensive risk metrics and drawdown analysis - **Comparative Value**: Clear insights vs baseline strategies --- ## 🛡️ Agent C: Risk Management Section ### Scope - **Primary Responsibility**: Risk analysis, portfolio optimization, and safety validation - **Core Tools**: `monte_carlo_*`, `risk_report`, `risk_analyze_portfolio`, `var_calculation`, `stress_test` - **Deliverables**: - Comprehensive risk assessment reports - Portfolio optimization recommendations - Risk-adjusted performance metrics - Stress testing results and scenario analysis - Monte Carlo simulation outcomes ### Integration Points - **Receives From**: Agent B (performance data for risk analysis) - **Provides To**: Agent A (risk constraints for strategy development) - **Shared Outputs**: Risk reports, portfolio recommendations, safety guidelines ### Success Criteria - **Risk Management**: Maximum drawdown < 15%, VaR within acceptable limits - **Portfolio Optimization**: Improved risk-adjusted returns vs baseline - **Stress Resilience**: Strategy performs well under adverse conditions - **Compliance**: All risk parameters within predefined guidelines --- ## 🔄 Parallel Execution Guidelines ### File Ownership Boundaries - **Agent A**: `~/source/jesse-mcp/strategies/*/development/` - **Agent B**: `~/source/jesse-mcp/strategies/*/backtesting/` - **Agent C**: `~/source/jesse-mcp/strategies/*/risk-analysis/` - **Shared State**: `~/source/jesse-mcp/shared-state/` ### Tool Exclusivity - **No Overlap**: Each tool assigned to single primary agent - **Clear Interfaces**: Defined input/output contracts between agents - **Dependency Management**: Required inputs must be available before execution ### Data Flow Directionality - **A → B**: Strategy files and testing parameters - **B → C**: Performance metrics and analysis results - **C → A**: Risk constraints and safety guidelines - **C → All**: Portfolio-wide risk assessments ### Communication Protocols - **Status Updates**: File-based progress indicators - **Data Transfer**: JSON schemas with validation - **Error Handling**: Clear escalation and recovery procedures - **Version Control**: Schema evolution and backward compatibility --- ## 📅 Development Timeline ### Phase 1: Foundation (Days 1-2) - [ ] Directory structure creation and shared state setup - [ ] Agent profile system implementation - [ ] Task mapping and detection logic - [ ] Error handling and SUSPENDED framework ### Phase 2: Core Implementation (Days 3-4) - [ ] Agent A: Strategy development capabilities - [ ] Agent B: Backtesting and analysis framework - [ ] Agent C: Risk management and safety validation - [ ] Inter-agent communication protocols ### Phase 3: Integration (Days 5-6) - [ ] Parallel execution orchestration - [ ] Centralized PRD template completion - [ ] Cross-agent workflow testing - [ ] Performance optimization and load balancing ### Phase 4: Validation & Deployment (Days 7-8) - [ ] End-to-end system testing - [ ] Documentation and best practices guide - [ ] Production deployment and monitoring - [ ] Training materials and usage examples --- ## ⚙️ Configuration ### Global Settings - **Default Risk Profile**: moderate (2% portfolio risk, 3x leverage) - **Parallel Execution**: Maximum 3 concurrent agents - **Shared State Location**: `~/source/jesse-mcp/shared-state/` - **Backup Strategy**: Daily snapshots of all strategy data ### Agent-Specific Settings - **Agent A Temperature**: 0.2 (precise execution) - **Agent B Temperature**: 0.7 (analytical iteration) - **Agent C Temperature**: 0.3 (conservative analysis) - **Timeout Settings**: 30 minutes per agent task - **Retry Logic**: 3 attempts with exponential backoff ### Integration Configuration - **Schema Validation**: Strict JSON schema enforcement - **Version Compatibility**: Automatic migration between versions - **Error Recovery**: Automatic retry and fallback mechanisms - **Performance Monitoring**: Real-time agent status tracking --- ## 📝 NOTES ### Initial Development Notes - **2025-12-11**: Centralized PRD structure established for parallel development - **2025-12-11**: Agent specialization defined with clear boundaries and interfaces - **2025-12-11**: Task mapping logic created for automatic profile selection - **2025-12-11**: File-based communication protocols selected for reliability ### Architecture Decisions - **Parallel over Sequential**: Chosen for maximum development efficiency - **File-Based Communication**: Selected over complex IPC for simplicity and reliability - **Schema-First Approach**: JSON schemas defined before implementation - **Error Resilience**: SUSPENDED.md symlinks for clear status signaling ### Lessons Learned - **Clear Boundaries**: Essential for preventing agent conflicts and resource contention - **Defined Interfaces**: Critical for reliable inter-agent communication - **Version Control**: Necessary for managing evolving agent capabilities - **Testing Framework**: Required for validating parallel execution workflows