inity13/decisionmatrix-mcp

Overview

Deterministic multi-criteria decision analysis (MCDA) for AI agents over MCP. Scores, ranks, and explains options against weighted criteria using weighted-sum, weighted-product, and TOPSIS methods with exact decimal arithmetic. Available as a live remote server or self-hosted.

Best for

Best for
Developers building AI agents that need transparent, reproducible multi-criteria ranking over MCP.

Use cases

• Rank candidate solutions against weighted business or technical criteria
• Provide transparent, reproducible decision explanations for agent workflows
• Compare alternatives with TOPSIS when trade-offs need distance-based scoring

How to use

Install

npx -y decisionmatrix-mcp

Tools exposed

• create_decision
• score_options
• sensitivity_analysis
• compare_two
• list_methods
• health_check
• weighted_product

Tested with

Claude Desktop, Cursor, VS Code

Example client config

{ "mcpServers": { "decisionmatrix": {\n    "type": "http", "url": "https://decisionmatrix-mcp.pages.dev/mcp" } } }

Notes

Deterministic multi-criteria decision analysis (MCDA) for AI agents over MCP. Scores, ranks, and explains options against weighted criteria using weighted-sum, weighted-product, and TOPSIS methods with exact decimal arithmetic. Available as a live remote server or self-hosted.

0 stars on GitHub. Last updated 2026-08-12. Licensed MIT.

Use cases

• Rank candidate solutions against weighted business or technical criteria
• Provide transparent, reproducible decision explanations for agent workflows
• Compare alternatives with TOPSIS when trade-offs need distance-based scoring

Pros

• Deterministic and exact decimal arithmetic avoids floating-point inconsistencies
• Multiple MCDA methods (weighted-sum, weighted-product, TOPSIS) cover different decision styles
• Flexible deployment: remote server or self-hosted

Cons

• No stars or community traction yet, so maturity and support are unproven
• Requires MCP integration, limiting use to MCP-compatible clients
• Criteria weighting must be supplied manually; no built-in weight elicitation

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.