Increasing test coverage in your company with GitHub Copilot

Understand features, enable developers, and measure Copilot's impact.

Quem pode usar esse recurso?

GitHub Copilot Business or GitHub Copilot Enterprise

Neste artigo

O guia é inspirado pelo ESSP (Guia estratégico do sistema de engenharia) do GitHub, que recomenda estratégias e métricas para impulsionar melhorias em sistemas de engenharia.

Se você estiver iniciando uma distribuição do Copilot, recomendamos definir suas metas, planejar a distribuição adequadamente e comunicar as metas com clareza à equipe. Confira Alcançar as metas de engenharia da sua empresa com o GitHub Copilot.

1. Identify barriers to success

A primeira etapa recomendada pelo ESSP é desenvolver uma compreensão clara dos obstáculos que impedem melhorias na empresa. Ao entender sua linha de base atual, seu estado futuro desejado e as barreiras que impedem o progresso, você pode garantir alterações direcionadas e eficazes.

Many software teams face persistent challenges in maintaining high-quality code due to low unit test coverage. In fast-paced development environments, writing tests is often seen as time-consuming or non-essential, especially when teams are under pressure to deliver features quickly.

As a result, critical bugs can be discovered late in the development lifecycle, often in staging or production environments.

This leads to a chain of negative outcomes:

  • Higher bug rates and customer-reported issues
  • Increased cost of fixing bugs after deployment
  • Reduced developer confidence in the stability of their code
  • Slower release cycles due to reactive debugging and patching

In legacy systems, test coverage can be even harder to address because of complex dependencies or poorly documented code. Developers may lack familiarity with older codebases or with testing frameworks in general, further compounding the problem.

Improving test coverage is a recognized best practice, but it requires time and expertise that many teams struggle to allocate.

2. Evaluate your options

A próxima etapa é avaliar e concordar com soluções para resolver as barreiras identificadas na etapa um. Neste guia, vamos focar o impacto que o GitHub Copilot pode ter na meta identificada. Lembre-se de que distribuições bem-sucedidas de uma nova ferramenta também exigem alterações na cultura e nos processos.

Você executará testes de novas ferramentas e processos com grupos piloto para coletar comentários e medir o sucesso. Para recursos de treinamento e métricas a serem usados durante as avaliações, leia as seções 3. Implementar alterações e Métricas a observar.

Sign up for Copilot

How Copilot can help

GitHub Copilot can significantly accelerate and simplify the process of writing unit tests. By understanding the surrounding code and context, Copilot can suggest test functions that match the structure and logic of the code being tested.

Copilot's capabilities are useful across multiple scenarios:

  • As developers write new functions, Copilot can automatically suggest corresponding test cases inline.
  • When refactoring legacy code, Copilot can help generate test scaffolding to prevent regressions.
  • For untested modules, developers can prompt Copilot to generate meaningful test cases, even when test coverage is missing or inconsistent.

By making unit testing easier, faster, and less manual, Copilot reduces the friction that can lead to gaps in test coverage, and helps teams adopt a quality-first mindset.

Use cases

  • Inline test generation: Developers can ask Copilot to generate tests for a specific function or module without switching context.
  • Better edge case coverage: By prompting Copilot for edge scenarios (such as null inputs, empty lists, or invalid states), developers can quickly cover more branches of logic.
  • Accelerated onboarding: New team members can use Copilot to understand how a function is expected to behave by looking at the generated test cases.
  • Assistance with CI/CD: Copilot can suggest how to integrate tests into your build pipeline, ensuring that coverage improvements directly support quality gates.

Cultural considerations

Além da distribuição do GitHub Copilot, você também deve abordar fatores sociais ou culturais que possam impedir que você atinja suas metas.

Os exemplos a seguir são extraídos da seção "Anti-Patterns" no ESSP.

  • Teams might rely on manual testing or insufficient automated testing. This could be caused by resource constraints for automation or a lack of experience with modern test tools.
  • Teams might wait too long to release, deploying large batches of code at once, which makes bugs and regressions harder to detect. This could be caused by a lack of CI/CD pipeline maturity, strict compliance requirements, or long review cycles between PR and deployment.

3. Implement changes

Quando você identificar a abordagem certa para superar suas barreiras, dimensionará as soluções identificadas. Para uma distribuição bem-sucedida de uma nova ferramenta ou processo, é importante atribuir a propriedade a cada parte da distribuição, comunicar-se de modo transparente sobre suas metas, fornecer treinamento eficaz e medir seus resultados.

Esta seção apresenta exemplos cenários, práticas recomendadas e recursos para desenvolvedores. É recomendável usar esta seção para planejar as sessões de comunicação e treinamento para ajudar os funcionários a usar o Copilot de um modo alinhado à meta.

Generate tests inline

  1. In VS Code, select the function you want to test and prompt Copilot: Generate a unit test for this code.
  2. Copilot generates a test inline or in a separate test file, depending on the language and structure.
  3. Review, refine, and accept the suggestion.

Cover edge cases

  1. After writing a test, ask Copilot: What are some edge cases I should test for this function?

    Or: Write test cases for when the input is null or empty.

  2. Copilot suggests additional test cases to cover boundary conditions.

  3. Review the tests and incorporate them into your test suite.

Understand new code

  1. Select a legacy function and ask Copilot: Explain what this function does and generate a test to validate it.
  2. Copilot explains the function's purpose and suggests corresponding test cases.
  3. Look at the test cases to understand the expected behavior and quickly build context.

Get assistance with CI/CD

  1. Review your test cases and commit them to the codebase.
  2. Ask Copilot: Where should I place this test if I want it to run in CI?
  3. Based on the structure of the codebase, Copilot will suggest where to place test files and how to update pipeline configurations.

Best practices for developers

Developers should:

  • Use descriptive comments or prompts when chatting with Copilot. For example: Generate unit tests for a function that calculates discounts based on user type and purchase amount.
  • Use Copilot to explore logic coverage. For example: What branches or conditions does this function have that should be tested?
  • Explore different prompt techniques and compare results from different AI models.

Developers should not:

  • Accept generated tests without reviewing logic. Make sure the tests reflect actual requirements and handle realistic inputs and outputs.
  • Skip asserting edge behavior. If you only test "happy paths," you risk missing regressions.
  • Rely on Copilot to guess undocumented business rules. Always provide context through prompts or comments.
  • Treat Copilot as a substitute for human code reviews. Copilot accelerates the process, but you still need to apply engineering judgment.

Resources for developers

Metrics to watch

Para analisar avaliações de novas ferramentas e garantir que suas distribuições completas estejam fornecendo melhorias consistentes, monitore os resultados e faça ajustes quando necessário. Em geral, recomendamos considerar as principais zonas de qualidade, velocidade e satisfação do desenvolvedor e como essas zonas se reúnem para contribuir com os resultados dos negócios.

Aqui estão algumas métricas que recomendamos analisar para avaliar o impacto do Copilot nessa meta específica.

  • Test coverage: Track increases in line and branch coverage after Copilot adoption. If possible, look at test coverage reports from your CI pipelines.
  • Bug rate after deployment: Fewer bugs should be reported in production environments.
  • Developer confidence: Use surveys or retrospectives to assess how confident developers feel shipping new code.
  • Time to write tests: Measure reduction in time spent creating unit tests.