UNPKG

@spaik/mcp-server-roi

Version:

MCP server for AI ROI prediction and tracking with Monte Carlo simulations

141 lines (118 loc) 5.93 kB
# Changelog All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). ## [1.3.1] - 2025-07-09 ### Fixed - **Critical Bug Fix**: Fixed "Cannot read properties of undefined (reading 'industry')" error in predict_roi tool - Added proper Zod schema validation for input arguments in index.ts - Added safety check in predict-roi.ts to validate project data exists before accessing properties - Applied same validation pattern to compare_projects tool for consistency - **Input Validation**: Tools now properly parse and validate input arguments before processing - Prevents runtime errors from malformed input - Provides clear validation error messages to users - Ensures type safety throughout tool execution ### Improved - Enhanced error handling with more descriptive validation messages - Better consistency between predict_roi and compare_projects tools ## [1.3.0] - 2025-07-03 ### Breaking Changes - **Removed quick-assessment tool**: This tool has been completely removed from the codebase - **PERPLEXITY_API_KEY is now mandatory**: Required for Dutch market validation in all tools ### Added - **Dutch Market Validation**: All ROI predictions and project comparisons now include mandatory Dutch market validation - New `DutchBenchmarkValidator` service integrates with Perplexity API - Automatic validation of input values against Dutch industry averages - Intelligent adjustment of unrealistic values (e.g., savings > 3x industry average) - Detailed rationale provided for all adjustments - **Enhanced Validation Results**: Both predict-roi and compare-projects tools now include: - `dutchValidation` section with adjustment details - Market-specific insights and trends - Citations from authoritative Dutch sources - Confidence scores based on market alignment ### Changed - predict-roi tool: Perplexity validation is now the mandatory first step - compare-projects tool: Each project undergoes individual Dutch market validation - Configuration validation now requires PERPLEXITY_API_KEY - Tool descriptions updated to mention Dutch market focus ### Removed - quick-assessment tool and all related code - QuickUseCase schema and associated methods - All quick assessment functionality from service files ### Security - PERPLEXITY_API_KEY validation added to startup checks - Enhanced error handling for API failures ## [1.2.0] - 2025-07-01 ### Added - **Natural Language Processing**: All tools now support natural language input - Universal NLP service for intent detection and entity extraction - Context awareness maintains conversation history - Support for relative values ("reduce by half", "triple the output") - **Smart Defaults**: 80% reduction in required input fields - Industry-specific implementation cost defaults - Intelligent timeline estimation - Automatic use case generation based on project type - **Flexible Input Formats**: - Accept currency strings: "$50k", "$1.5M", "100 thousand dollars" - Accept percentages: "85%", "90 percent", "half" - Accept time periods: "6 months", "2 years", "quarterly" - Fuzzy matching for enums: "tech" → "technology", "fin" → "financial_services" - **Enhanced Error Handling**: - User-friendly error messages with examples - "Did you mean?" suggestions for common mistakes - Auto-correction for simple errors - Field-specific guidance and constraints - **New Utility Tools**: - `get_examples`: Retrieve relevant usage examples for any tool - `help`: Interactive help system with tool recommendations - **LLM Optimizations**: - Self-documenting tools with embedded examples - Structured responses with confidence indicators - Next steps and alternative approaches in responses - Comprehensive prompt engineering guide ### Changed - Enhanced all tool descriptions with capabilities and examples - Improved response structure with `next_steps` and `confidence_indicators` - Updated error responses to include recovery suggestions - Simplified input schemas while maintaining backward compatibility ### Fixed - Error messages now provide actionable guidance - Enum validation accepts common variations - Numeric parsing handles various formats ## [1.1.1] - 2025-06-30 ### Fixed - Database connection handling improvements - Performance optimizations for Monte Carlo simulations - Error message clarity ## [1.1.0] - 2025-06-28 ### Added - Enhanced ML comparison engine - Improved industry benchmark integration - Voice output support for accessibility ### Changed - Optimized worker pool performance - Enhanced response transformer logic ## [1.0.0] - 2025-06-24 ### Added - Initial release with full MCP implementation - Three core tools: predict_roi, compare_projects, quick_assessment - Monte Carlo simulation with parallel processing - Real-time industry benchmarks via Perplexity API - ML-powered project comparison - Natural language support for quick_assessment - Comprehensive transaction management - Atomic operations with rollback support - Retry logic with exponential backoff - Financial calculators (NPV, IRR, Payback Period) - 5-year ROI projections - Industry-specific calculations for 7 sectors - Supabase PostgreSQL integration - Worker thread pool for CPU-intensive tasks - Structured logging with context - Graceful API degradation [1.3.1]: https://github.com/SPAIK-io/mcp-server-roi/compare/v1.3.0...v1.3.1 [1.3.0]: https://github.com/SPAIK-io/mcp-server-roi/compare/v1.2.0...v1.3.0 [1.2.0]: https://github.com/SPAIK-io/mcp-server-roi/compare/v1.1.1...v1.2.0 [1.1.1]: https://github.com/SPAIK-io/mcp-server-roi/compare/v1.1.0...v1.1.1 [1.1.0]: https://github.com/SPAIK-io/mcp-server-roi/compare/v1.0.0...v1.1.0 [1.0.0]: https://github.com/SPAIK-io/mcp-server-roi/releases/tag/v1.0.0