Three engines, one server

sim_* · simulate

ScenarioSim: project SaaS growth, pricing, churn, runway, unit economics forward in time; sensitivity + break-even.

decide_* · decide

DecisionMatrix: rank options against weighted criteria (weighted-sum, weighted-product, TOPSIS) with full breakdowns.

calc_* · compute

PrecisionCalc: exact SaaS/finance metrics, currency, business days, NPV, IRR, loan amortization, depreciation.

deterministic

Every number runs through 40-digit decimal math. Identical inputs → byte-identical output across all three.

one key, one quota

A single API key and daily quota spans the whole stack — cheaper and simpler than three separate servers.

profiles

Load only what you need: ?profile=finance, decision, or simulation to cut tool-selection noise.

Composite tools — better than the sum

These chain the engines to do reasoning no single server can. Deterministic end to end.

Tool	Pipeline	What it does
`plan_to_valuation`	simulate → compute	Project a scenario, then value its cash-flow line: NPV, IRR, undiscounted total.
`evaluate_options_with_scenarios`	simulate → decide	Project each option as a scenario, then rank the outcomes against weighted criteria.
`stress_test_decision`	simulate × decide	Stress a scenario assumption across every option; report how often the winner holds.

Connect your client

Cursor

Claude Desktop

VS Code

curl

~/.cursor/mcp.json

json

{
  "mcpServers": {
    "agentstack": {
      "url": "https://agentstack-mcp.pages.dev/mcp"
    }
  }
}

Focused subset: use .../mcp?profile=finance (or decision / simulation).

claude_desktop_config.json — via mcp-remote bridge

json

{
  "mcpServers": {
    "agentstack": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://agentstack-mcp.pages.dev/mcp"]
    }
  }
}

.vscode/mcp.json

json

{
  "servers": {
    "agentstack": {
      "type": "http",
      "url": "https://agentstack-mcp.pages.dev/mcp"
    }
  }
}

bash

curl -s https://agentstack-mcp.pages.dev/mcp \
  -H 'content-type: application/json' \
  -H 'accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{
        "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},
                      {"metric":"total_churned_customers","weight":1,"direction":"cost"}]}}}'

Pricing

Start free. One key covers all three products and the composites — about half the cost of subscribing to them separately. Pay by card, get a key instantly, send it as X-API-Key. Cancel anytime.

Free

20 calls / day
All 24 tools · all composites
All profiles
No API key needed
Shared IP-based quota

Use free endpoint

For evaluation & low-volume testing.

Starter Best value

$19/mo

8,000 calls / day
All three products + composites
One key · one quota
vs ~$36/mo bought separately
Billing portal

Get Starter →

The whole stack for one production agent.

Pro

$59/mo

80,000 calls / day
Everything in Starter
Highest priority
Billing portal
Email support

Get Pro →

For fleets & high-volume workloads.

Secure checkout by Stripe · promo codes supported · manage anytime at the billing portal. Prices in USD.

Prefer one domain only? The three products also run standalone: ScenarioSim · DecisionMatrix · PrecisionCalc. Self-host AgentStack (MIT) for unlimited calls.