Hcode
IA para Devs

GitHub Copilot Avançado

2 - 4 Hours 4 Modules

Overview

Main Theme: AI for Developers

Duration: 2 - 4 Hours

Level: Advanced

Target Audience: Software Developers, Data Engineers and Systems Architects

ABOUT THE TRAINING

Mastering GitHub Copilot goes far beyond accepting line code suggestions. This training focuses on the strategic potential of the tool: from translating complex legacy systems to implementing unit tests that ensure application security. The goal is to transform Copilot into a career accelerator, allowing developers to focus on business logic while AI handles technical implementation and documentation.

Objective

Empowering professionals to use advanced context engineering strategies to solve architectural issues, modernize legacy codebases, and ensure high-quality software delivery with a reduction in routine coding time.

TRAINING MODULES

Módulo 01

Context Engineering and Workspace

Learning how to "teach" the project architecture to Copilot. Techniques for selecting the right files and contexts in the chat so that suggestions are consistent with the project design pattern, avoiding isolated solutions that do not fit into the overall structure.

Módulo 02

AI-Assisted TDD (Test-Driven Development)

Implementation of the methodology of writing the test first. The student will learn to describe the expected behavior and request that Copilot generates both complex unit tests and the functional code that satisfies them, reversing the traditional flow and increasing test coverage.

Módulo 03

Refactoring, Design Patterns, and Performance

Identification of "code smells" and technical debt. The student will learn to request that Copilot rewrites code snippets following modern design patterns and optimizes algorithm performance, transforming legacy code into modular, clean, and high-performance solutions.

Módulo 04

Prompt Templates and Documentation Automation

Advanced use of template patterns (such as those seen in frameworks like LangChain). The student will learn to create structured prompts with multiple variables to generate automatic technical documentation and maintain consistent code standards across the development team.

Configurando um agente GitHub Copilot

agente_desenvolvedor.agent.md
---
  description: 'Especialista em Engenharia de Software e IA, focado em refatoração, padrões de projeto (Design Patterns) e otimização de performance em Python e Node.js'
  Configurar as Ferramentas...
  tools: ['vscode', 'execute', 'read/readFile', 'edit/createDirectory', 'search', 'agent', 'pylance-mcp-server/*', 'ms-python.python/getPythonEnvironmentInfo', 'ms-python.python/getPythonExecutableCommand', 'ms-python.python/installPythonPackage', 'ms-python.python/configurePythonEnvironment']
---

Define what this custom agent accomplishes for the user, when to use it, and the edges it won't cross. Specify its ideal inputs/outputs, the tools it may call, and how it reports progress or asks for help.

Padrões de Prompt Templates em LangChain

Com base no notebook que você está analisando, aqui estão os principais padrões:

1. PromptTemplate Básico

Cria um template com variáveis entre chaves {}:

Python
prompt_template = PromptTemplate.from_template('''
Responda a seguinte pergunta do usuário:
{pergunta}
''')
prompt_template.format(pergunta='O que é um buraco negro?')

2. PromptTemplate com Múltiplas Variáveis

Define várias variáveis que serão preenchidas depois:

Python
prompt_template = PromptTemplate.from_template('''
Responda a pergunta em até {n_palavras} palavras:
{pergunta}
''')

3. Partial Variables (Variáveis Pré-definidas)

Fixa algumas variáveis para reutilizar sem especificá-las sempre:

Python
prompt_template = PromptTemplate.from_template(
    'Responda em até {n_palavras} palavras: {pergunta}',
    partial_variables={'n_palavras': 10}
)

# Agora só precisa passar 'pergunta'

Get in touch with our specialists