io.github.inity13/scenariosim-mcp

Measured conclusion: Deep-tested. L2 outcome: tools_ok.

Deterministic what-if & scenario simulation for AI agents: projections, sensitivity & break-even.

Add to your agent

Copy this into your MCP client config:

{
  "mcpServers": {
    "scenariosim-mcp": {
      "url": "https://scenariosim-mcp.pages.dev/mcp",
      "transport": "streamable-http"
    }
  }
}

Measured tools (6)

• run_scenario

  Main simulation tool. Run a deterministic what-if projection from a pre-built template (saas_growth, pricing_change, churn_impact, cost_reduction, hiring_plan, cash_runway, unit_economics, marketing_funnel, compound_growth) OR a free-form 'metrics' model. Returns period-by-period projections, headline key_results, the exact assumptions used (with defaults filled in), the methodology, notes, and a plain-language explanation. Pass 'template' + 'inputs' (assumptions), plus optional 'horizon' and 'period_label'. 100% deterministic (40-digit decimal math).

  Parameters:

  • inputs — object — Scenario assumptions as {name: value}. Which keys are valid depends on the template (call list_templates). Unlisted keys fall back to documented defaults; unknown keys are ignored and reported in notes. You may also pass assumptions at the top level.
  • horizon — integer — Number of periods to project forward (1..1200). Default depends on template (usually 12).
  • metrics — array — For a CUSTOM free-form scenario (template omitted or 'custom'): a list of independently-growing metrics. Each: {name, start, growth_rate (per period, default 0), mode: 'compound' (default) | 'linear'}.
  • template — string — Pre-built scenario template id: saas_growth, pricing_change, churn_impact, cost_reduction, hiring_plan, cash_runway, unit_economics, marketing_funnel, compound_growth (aliases like 'saas','pricing','runway','ltv' also resolve). Omit (or use 'custom') to run a free-form 'metrics' projection.
  • period_label — string — Label for each period; also sets annualization (periods/year).

  Hints: read-only not declared · destructive not declared · idempotent not declared · open-world not declared

• sensitivity_analysis

  Vary one or more input assumptions and show the impact on a target output metric (one-at-a-time sensitivity). Provide 'template', the input to sweep via 'variable' (or 'variables' array), and 'target_metric' (defaults to the template's primary output). Control the sweep with 'variation' (fractional +/- around the baseline, default 0.2), 'steps' (default 5), or explicit 'values' / 'min'+'max'. Returns per-variable sweeps, an elasticity estimate, the output range, and a ranking of the most influential inputs.

  Parameters:

  • template (required) — string — Pre-built scenario template id: saas_growth, pricing_change, churn_impact, cost_reduction, hiring_plan, cash_runway, unit_economics, marketing_funnel, compound_growth (aliases like 'saas','pricing','runway','ltv' also resolve). Omit (or use 'custom') to run a free-form 'metrics' projection.
  • max — number — Sweep upper bound (with 'min').
  • min — number — Sweep lower bound (with 'max').
  • steps — integer — Number of sweep points per variable (2-200).
  • inputs — object — Scenario assumptions as {name: value}. Which keys are valid depends on the template (call list_templates). Unlisted keys fall back to documented defaults; unknown keys are ignored and reported in notes. You may also pass assumptions at the top level.
  • values — array — Explicit sweep values for a single 'variable'.
  • horizon — integer — Number of periods to project forward (1..1200). Default depends on template (usually 12).
  • variable — string — A single input name to sweep.
  • variables — array — Multiple inputs to sweep (one at a time). Each: {name, variation?|values?|min?+max?, steps?}.
  • variation — number — Fractional sweep around the baseline (0<v<=1). 0.2 = +/-20%.
  • period_label — string — Label for each period; also sets annualization (periods/year).
  • target_metric — string — Output metric to track (see a template's 'outputs' via list_templates). Defaults to the template's primary output.

  Hints: read-only not declared · destructive not declared · idempotent not declared · open-world not declared

• break_even

  Solve for the input value required to make an output metric hit a target value (deterministic bisection root-finding). Provide 'template', 'solve_for' (the input to solve), 'target_metric' (defaults to the primary output), and 'target_value'. Optionally pass 'bounds' [low, high] to constrain the search. Returns the required input value, the change from baseline, the achieved metric, and the residual. Assumes the metric is monotonic in the solved input over the range.

  Parameters:

  • template (required) — string — Pre-built scenario template id: saas_growth, pricing_change, churn_impact, cost_reduction, hiring_plan, cash_runway, unit_economics, marketing_funnel, compound_growth (aliases like 'saas','pricing','runway','ltv' also resolve). Omit (or use 'custom') to run a free-form 'metrics' projection.
  • solve_for (required) — string — Name of the input variable to solve for.
  • target_value (required) — number — The value the target_metric should reach.
  • bounds — array — Optional [low, high] search range for the solved input. Auto-derived + expanded if omitted.
  • inputs — object — Scenario assumptions as {name: value}. Which keys are valid depends on the template (call list_templates). Unlisted keys fall back to documented defaults; unknown keys are ignored and reported in notes. You may also pass assumptions at the top level.
  • horizon — integer — Number of periods to project forward (1..1200). Default depends on template (usually 12).
  • period_label — string — Label for each period; also sets annualization (periods/year).
  • target_metric — string — Output metric to hit (defaults to the template's primary output).

  Hints: read-only not declared · destructive not declared · idempotent not declared · open-world not declared

• compare_scenarios

  Run 2-3 scenarios and compare their key_results side by side, with deltas against the first (baseline) scenario. Provide a 'scenarios' array where each entry is {name?, template, inputs} (each may set its own horizon, or pass a shared top-level 'horizon'). Optionally rank on 'compare_metric' with 'goal' ('max' default | 'min') to pick a winner, and set include_projections:true to also return per-period series.

  Parameters:

  • scenarios (required) — array — 2-3 scenarios to compare. Each: {name?, template, inputs, horizon?, period_label?} or {name?, metrics:[...]} for a custom model.
  • goal — string — Whether higher (max) or lower (min) is better for compare_metric.
  • horizon — integer — Optional shared horizon applied to scenarios that don't set their own.
  • compare_metric — string — Metric to rank scenarios on (optional).
  • include_projections — boolean — Include each scenario's full per-period projections.

  Hints: read-only not declared · destructive not declared · idempotent not declared · open-world not declared

• list_templates

  Discovery tool: list every pre-built scenario template (id, label, category, description, primary output, documented inputs with defaults/units, and available output metrics), plus how to run a custom free-form scenario and the supported period labels. No required parameters.

  Hints: read-only not declared · destructive not declared · idempotent not declared · open-world not declared

• health_check

  Server health, version, and capabilities (tools, templates, period labels, max horizon). No parameters.

  Hints: read-only not declared · destructive not declared · idempotent not declared · open-world not declared

Repository

Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.