Getting Started with Vercel Eve: Build Your First Agent Using eve init

Vercel Eve requires Node.js 24 or newer. Verify the version with node -v and, if necessary, switch using nvm install 24 && nvm use 24. A Vercel AI Gateway API key is also required.

Initialize the project

Run the following command inside the tutorial repository: npx eve@latest init 01-first-agent The command performs four actions:

Creates the 01-first-agent/ directory.

Writes a minimal agent.ts file.

Installs project dependencies.

Generates the built‑in Eve channel.

Project structure

example/01-first-agent/
  package.json
  tsconfig.json
  .env.example
  agent/
    agent.ts
    instructions.md
    channels/
      eve.ts

Configure the first agent

Edit agent/agent.ts to define the model and its context window. The model ID can be overridden with the EVE_GATEWAY_MODEL_ID environment variable.

import { defineAgent } from "eve";

const defaultGatewayModelId = "minimax/minimax-m3";

export default defineAgent({
  model: process.env.EVE_GATEWAY_MODEL_ID ?? defaultGatewayModelId,
  modelContextWindowTokens: 200_000,
});

Write always‑on instructions

The agent/instructions.md file defines the agent’s stable identity, responsibilities, limits, and output style. Example content for the SpringForAll content assistant:

# SpringForAll Content Assistant

You are the first content operations assistant for the SpringForAll community.
Your mission is to help maintain a Chinese technical community for Java and Spring developers.

## Responsibilities
- Explain what you can do for SpringForAll content operations.
- Help brainstorm Java, Spring, Spring AI, Spring Cloud, JVM, backend engineering, and developer tooling topics.
- Turn vague topic ideas into practical article angles, outlines, and follow‑up questions.
- Keep answers practical and useful for engineers.
- Ask for human confirmation before treating any content as publish‑ready.

## Current limits
- You do not have skills yet.
- You do not have subagents yet.
- You do not have custom tools yet.
- You do not publish content automatically.
- You should not claim to have checked live sources unless the user provides them.

## Output style
- Write in concise Chinese by default.
- Use Markdown when structure helps.

Configure environment variables

Create .env.example with the required variables and copy it to .env for local development:

EVE_GATEWAY_MODEL_ID=minimax/minimax-m3
AI_GATEWAY_API_KEY=

Copy the template: cp .env.example .env Fill AI_GATEWAY_API_KEY with your Vercel AI Gateway key. Do not commit .env to the repository; keep only .env.example.

Validate the agent discovery

Run the info command to ensure Eve discovers the agent/ directory, recognises instructions.md, and reports zero skills (as intended for this minimal demo):

npm exec -- eve info

Application
App Root      .../example/01-first-agent
Agent Root    .../example/01-first-agent/agent
Layout        nested
Compile       ready
Diagnostics   0 errors, 0 warnings
Instructions  instructions.md
Skills        0 skills

Build and run the agent

Build the project to verify compilation: npm run build Start development mode, which launches Eve’s interactive TUI chat: npm run dev In the chat, type a query such as: 你是谁？你能帮 SpringForAll 做什么？ The expected answer should be in Chinese, acknowledge the agent’s role as a SpringForAll content assistant, list the supported tasks (topic brainstorming, outline suggestions, etc.), and must not claim automatic publishing or live web search capabilities.

What the minimal agent provides

A single main agent.

An always‑on instructions.md defining identity and limits.

Model access via Vercel AI Gateway.

Interactive CLI chat through eve dev.

Project‑structure verification with eve info and eve build.

Intentionally omitted features

Custom providers.

Skills, sub‑agents, sandbox, tools, schedules, and evals.

Repository locations

Example project:

https://github.com/dyc87112/vercel-eve-content-team-tutorial/tree/main/example/01-first-agent

Full tutorial repository:

https://github.com/dyc87112/vercel-eve-content-team-tutorial