Building Your First Agent in Astreus
Learn to build your first AI agent with Astreus. A quick-start guide covering agent creation, task execution, and essential API methods.
Building AI agents doesn't need to be complicated. With Astreus, you can create a working agent in just a few lines of code. This tutorial covers the fundamentals using the real Astreus API.
## Quick Start
You can clone the complete example or start from scratch:
```bash
git clone https://github.com/astreus-ai/your-first-agent
cd your-first-agent
npm install
```
Or install just the package:
```bash
npm install @astreus-ai/astreus
```
## Environment Setup
Create a `.env` file with your configuration:
```env
OPENAI_API_KEY=sk-your-openai-api-key-here
DB_URL=sqlite://./astreus.db
```
The database URL can point to SQLite for local development or PostgreSQL for production.
## Creating Your First Agent
Here's the basic agent implementation:
```typescript
import { Agent } from '@astreus-ai/astreus';
const agent = await Agent.create({
name: 'MyFirstAgent',
model: 'gpt-4o',
systemPrompt: 'You are a helpful assistant.'
});
```
The `Agent.create()` method takes three core parameters: a name for identification, the LLM model to use, and a system prompt that defines the agent's behavior and personality.
## Working with Tasks
Tasks provide structure for agent interactions. Create a task with a prompt, then execute it:
```typescript
const task = await agent.createTask({
prompt: "Hello, introduce yourself"
});
const result = await agent.executeTask(task.id);
console.log(result.response);
```
The `createTask()` method generates a task object with an ID and status. The `executeTask()` method processes the task and returns the agent's response.
## Running Your Agent
If you cloned the repository:
```bash
npm run dev
```
For a standalone script:
```bash
npx tsx index.ts
```
## Core API Methods
Astreus provides three essential methods for basic agent operations:
- `Agent.create()` - Creates a new agent instance with configuration
- `agent.createTask()` - Generates a task with a prompt and optional metadata
- `agent.executeTask()` - Executes a task by ID and returns the response
These methods form the foundation for building more complex agent systems.
## Complete Example
Here's everything together:
```typescript
import { Agent } from '@astreus-ai/astreus';
const agent = await Agent.create({
name: 'MyFirstAgent',
model: 'gpt-4o',
systemPrompt: 'You are a helpful assistant.'
});
const task = await agent.createTask({
prompt: "Hello, introduce yourself"
});
const result = await agent.executeTask(task.id);
console.log(result.response);
```
This creates an agent, assigns it a task, and logs the response. The agent uses GPT-4 to process the prompt and generate an answer.
## What's Next
This is just the beginning. The Astreus framework supports more advanced features like memory, knowledge bases, vision capabilities, and multi-agent systems. Start with this foundation and expand as your needs grow.
## Source Code
The complete working example is available on GitHub: [astreus-ai/your-first-agent](https://github.com/astreus-ai/your-first-agent)
This experiment is written for Astreus v0.5.37. Please ensure you are using a compatible version.