Skip to content

templates.md

Latest commit

 

History

History
231 lines (181 loc) · 4.33 KB

templates.md

File metadata and controls

231 lines (181 loc) · 4.33 KB

Templates Guide

Agent-as-Code provides a powerful template system for quickly creating common types of AI agents. This guide explains how to use and create templates.

Available Templates

Template Description Use Case
chatbot Conversational agent Customer service, chat interfaces
sentiment Sentiment analysis Text analysis, feedback processing
summarizer Text summarization Document processing
translator Language translation Multi-language support
data-analyzer Data analysis Business intelligence
content-gen Content generation Marketing, creative writing

Using Templates

List Available Templates

agent init --list-templates

Create Agent from Template

agent init my-agent --template chatbot

Template Details

Chatbot Template

A conversational agent template with session management and API endpoints.

Features:

  • REST API with FastAPI
  • Session management
  • Health checks
  • Conversation history

Structure:

chatbot/
├── main.py              # Main agent code
├── requirements.txt     # Dependencies
└── README.md           # Documentation

API Endpoints:

# main.py
class ChatRequest(BaseModel):
    message: str
    session_id: Optional[str] = None

class ChatResponse(BaseModel):
    response: str
    timestamp: str

@app.post("/chat")
async def chat(request: ChatRequest):
    # Chat endpoint implementation

Sentiment Analysis Template

A template for analyzing text sentiment.

Features:

  • Sentiment analysis API
  • Confidence scores
  • Batch processing
  • Health monitoring

Structure:

sentiment/
├── main.py              # Main agent code
├── requirements.txt     # Dependencies
└── README.md           # Documentation

API Endpoints:

# main.py
class SentimentRequest(BaseModel):
    text: str
    include_confidence: Optional[bool] = True

class SentimentResponse(BaseModel):
    sentiment: str
    confidence: float
    timestamp: str

@app.post("/analyze")
async def analyze_sentiment(request: SentimentRequest):
    # Sentiment analysis implementation

Template Configuration

Each template includes:

1. agent.yaml Configuration

apiVersion: agent.dev/v1
kind: Agent
metadata:
  name: template-name
spec:
  runtime: python:3.11
  model:
    provider: openai
    name: gpt-4
  capabilities:
    - template-specific-capability

2. Dependencies

# requirements.txt
fastapi==0.104.0
uvicorn==0.24.0
pydantic==2.5.0

3. Documentation

# Template Name

Description of the template and its capabilities.

## Usage
1. Installation steps
2. Configuration guide
3. Example usage

Creating Custom Templates

Templates are embedded in the binary and follow a specific structure:

//go:embed chatbot/* sentiment/*
var templateFS embed.FS

Template Structure

template-name/
├── main.py              # Main implementation
├── requirements.txt     # Dependencies
├── README.md           # Documentation
└── template.yaml       # Template metadata

Template Metadata

# template.yaml
name: template-name
description: Template description
author: Your Name
version: 1.0.0
runtimes:
  - python
tags:
  - tag1
  - tag2

Template Best Practices

1. Code Organization

  • Clear file structure
  • Modular design
  • Type hints
  • Documentation

2. Error Handling

@app.exception_handler(Exception)
async def global_exception_handler(request, exc):
    return JSONResponse(
        status_code=500,
        content={"error": str(exc)}
    )

3. Health Checks

@app.get("/health")
async def health():
    return {
        "status": "healthy",
        "timestamp": datetime.now().isoformat()
    }

4. Configuration

class Settings(BaseSettings):
    model_provider: str
    model_name: str
    api_key: str
    
    class Config:
        env_file = ".env"

5. Logging

import logging

logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)

See Also