Is there an official TypeSafe MCP?

Community and vendor MCP packages evolve quickly. This page documents the integration pattern against our hosted decide API. Prefer packages you trust; point them at jv_live_ + /api/v1/decide on this host.

Jev MCP integration

MCP (Model Context Protocol) lets coding agents call tools. Wrapping Jev as an MCP tool gives the agent a typed decide surface — Choice / Score / Noul — without teaching it to invent JSON schemas every turn.

Agent skill guide →Try playground

What to expose

Start with one tool: decide(state, questions). Optionally add thin wrappers (triage_ticket, score_risk) that fix criteria maps so the agent cannot widen the label space by accident.

Auth

The MCP server should hold JEV_API_KEY (jv_live_…) server-side. Never paste the secret into the model context. Metering follows this host’s prepaid input-token balance.

Example tool sketch

Illustrative Node handler — adapt to your MCP SDK:

Example tool sketch

async function decideTool(args: { state: string; questions: Record<string, unknown> }) {
  const r = await fetch("https://jevtypesafe.org/api/v1/decide", {
    method: "POST",
    headers: {
      Authorization: `Bearer ${process.env.JEV_API_KEY}`,
      "Content-Type": "application/json",
    },
    body: JSON.stringify(args),
  });
  if (!r.ok) throw new Error(await r.text());
  return r.json();
}

FAQ

Is there an official TypeSafe MCP?: Community and vendor MCP packages evolve quickly. This page documents the integration pattern against our hosted decide API. Prefer packages you trust; point them at jv_live_ + /api/v1/decide on this host.