For TypeScript devs

Same API in TypeScript. Same Rust engine.

@agentmatic/core is napi-rs bindings to the same Rust core. Zod-typed tools, Vercel AI SDK adapter, works on Node 20 / Deno / Bun.

Install

npm install @agentmatic/core zod
# pnpm add @agentmatic/core zod
# bun add @agentmatic/core zod

Five-line agent

import { Agent, OpenAI, defineTool } from '@agentmatic/core';
import { z } from 'zod';

const calculator = defineTool({
  name: 'calculator',
  description: 'Evaluate a math expression',
  parameters: z.object({ expression: z.string() }),
  execute: async ({ expression }) => eval(expression).toString(),
});

const agent = Agent.builder('math-helper')
  .llm(new OpenAI())
  .tools([calculator])
  .build();

console.log(await agent.invoke('What is 42 * 17?'));

What's different from JS-only frameworks

• Same correctness as the Python SDK. Same engine, same semantics. No drift between SDK versions.
• Native binary speed. Prebuilt napi-rs binaries for darwin/linux/windows × x64/arm64. No node-gyp.
• Zod-first. Tools declare schemas with Zod; tool calls are validated at the boundary.
• Edge-runtime aware. Cloudflare Workers / Vercel Edge work — the runtime ships an isomorphic build (no native code in edge contexts; falls back to WASM).

What we ship for TS specifically

• Vercel AI SDK adapter — drop into streamText({ tools }).
• Express middleware — mount any agent as a route.
• OpenAI / Anthropic / Bedrock SDKs first-class.
• Full .d.ts exports; no any in public API.