Quickstart

Prerequisites

• Bun v1.1+ (or Node.js 20+)
• An API key from Anthropic, Gemini, LiteLLM or OpenAI

1. Create a Project

Using the CLI:

bunx rax init my-agent-app --template standard
cd my-agent-app
bun install

Or manually:

mkdir my-agent-app && cd my-agent-app
bun init -y
bun add reactive-agents effect

2. Set Up Environment

echo 'ANTHROPIC_API_KEY=sk-ant-...' > .env
echo 'GOOGLE_API_KEY=...' >> .env
echo 'TAVILY_API_KEY=tvly-...' >> .env
echo 'LITELLM_API_KEY=...' >> .env
echo 'OPENAI_API_KEY=...' >> .env

3. Build an Agent

Create src/agent.ts:

import { ReactiveAgents } from "reactive-agents";

async function main() {
  // await using disposes the agent automatically when the block exits
  await using agent = await ReactiveAgents.create()
    .withName("my-first-agent")
    .withProvider("anthropic")
    .withModel("claude-sonnet-4-20250514")
    .build();

  const result = await agent.run("What are the three laws of thermodynamics?");

  console.log("Output:", result.output);
  console.log("Duration:", result.metadata.duration, "ms");
  console.log("Steps:", result.metadata.stepsCount);
}

main();

Resource cleanup

Always dispose agents that use MCP servers or other subprocess-based tools — otherwise the process will hang on open pipes. Use await using for automatic cleanup, or runOnce() for one-shot scripts. See Resource Management for all three patterns.

4. Run It

bun run src/agent.ts

5. Add Capabilities

Enable memory, reasoning, and safety:

const agent = await ReactiveAgents.create()
  .withName("research-agent")
  .withProvider("anthropic")
  .withModel("claude-sonnet-4-20250514")
  .withMemory("1") // Tier 1: FTS5 search
  .withReasoning() // ReAct reasoning loop
  .withGuardrails() // Input safety checks
  .withCostTracking() // Budget enforcement
  .build();

What’s Next?

• Your First Agent — A deeper walkthrough
• Memory — How agent memory works
• Reasoning — Understanding reasoning strategies
• Architecture — The full layer system