AgentStack MCP: one deterministic reasoning stack for AI agents (simulate + decide + compute)

Developer inity13 released AgentStack MCP, a deterministic MCP server that unifies three previously separate engines for simulation, decision-making, and computation into a single endpoint. The server offers composite tools that chain these engines for end-to-end reasoning, such as evaluating options with scenarios and stress-testing decisions, with a free tier of 20 calls per day.

The fourth in a suite of deterministic MCP servers for AI agents — and the one that ties the first three together. Over the last stretch I shipped three focused, deterministic MCP servers: They're great on their own, but agents kept needing all three in the same task — and installing three servers, juggling three keys, and hand-gluing their outputs is friction. So here's AgentStack MCP : one endpoint, one key, all three — plus composite tools that chain them. { "mcpServers": { "agentstack": { "type": "http", "url": "https://agentstack-mcp.pages.dev/mcp" } } } Free tier: no key, 20 calls/day. The tools are namespaced so an agent always knows which engine it's calling: sim — ScenarioSim run, sensitivity, break-even, compare, templates decide — DecisionMatrix decide, score, sensitivity, compare two, methods calc — PrecisionCalc metrics, currency, NPV, IRR, loan, depreciation, … These chain the engines to do reasoning no single server can , deterministically end-to-end: evaluate options with scenarios simulate → decide — project each option as its own scenario, then rank the { "name": "evaluate options with scenarios", "arguments": { "template": "saas growth", "horizon": 12, "options": { "name": "Aggressive", "inputs": { "new customers per period": 60, "churn rate": 0.05 } }, { "name": "Lean", "inputs": { "new customers per period": 20, "churn rate": 0.02 } } , "criteria": { "metric": "ending mrr", "weight": 3, "direction": "benefit" }, { "metric": "total churned customers", "weight": 1, "direction": "cost" } } } plan to valuation simulate → compute — project a plan, then value its cash-flow line: NPV, IRR, undiscounted total. stress test decision simulate × decide — stress one scenario assumption across every option and report how often the chosen option survives robustness and where it flips. Bundling 24 tools risks drowning an agent's tool-selection. So the endpoint takes a ?profile= filter: https://agentstack-mcp.pages.dev/mcp?profile=finance finance → calc + plan to valuation decision → decide + the two decision composites simulation → sim + all composites all default → everything io.github.inity13/agentstack-mcp If your agents plan, choose, and do the numbers, give them one calculator that does all three — and never drifts. Feedback welcome.