Andrea Griffiths shares practical strategies for using GitHub Copilot on github.com. The article covers automating tasks, assigning AI agents, prototyping with Spark, switching between models, and optimizing your development workflow—without relying on your IDE.

How to Use GitHub Copilot on github.com: A Power User’s Guide

Authored by Andrea Griffiths

GitHub Copilot is best known as a code-completion assistant in your IDE, but there’s a powerful web-based side accessible right from github.com/copilot. This guide explores how to use Copilot’s AI features directly in your browser to streamline your workflow, automate routine tasks, and rapidly develop and prototype ideas.

Key Features and Strategies

1. File Issues from Screenshots

  • Drag-and-drop screenshots into Copilot chat.
  • Use simple prompts to auto-create issues, apply labels, and fill templates.
  • Copilot extracts information from images and suggests detailed issue content.
  • Tip: Let Copilot infer context for better, faster bug tracking.

2. Assign AI Agents to Tasks

  • Assign Copilot’s coding agent to issues via chat prompts (e.g., “Assign yourself to this issue and draft a fix.”).
  • The agent analyzes code, identifies root causes, and auto-generates pull requests.
  • Useful for bug fixes, documentation, and dependency updates.
  • Tip: Agents can follow up, run workflows, and work across issues.

3. Rapid Prototyping with Spark

  • Use GitHub Spark for live code scaffolding and real-time previews.
  • Great for practicing syntax, testing UI ideas, and sharing prototypes with your team.
  • Tip: Spark allows direct code editing and instant sharing.

4. Switching Between AI Models

  • Access multiple models: GPT-4.1, Claude Sonnet 4, and Opus 4.
  • Compare answers for the same prompt by switching models within a conversation thread.
  • Tip: Use model switching to get better explanations, refactors, or creative solutions.

5. Navigating Conversation Branches

  • Copilot groups multiple model responses as ‘branches’ under each message (like Git commit branches).
  • Review alternative solutions, explanations, or architectures side-by-side.

6. Combining Web and IDE Workflows

  • Use github.com for coordination, issue management, and prototyping.
  • Transition to VS Code for in-depth code implementation and debugging.
  • Example workflow: Start with project discussion on github.com, use Spark for prototyping, assign issues to team, complete development in the IDE.

Sample Daily Workflow

  1. Visit github.com/copilot.
  2. Check assigned pull requests and issues.
  3. Use chat to summarize project status.
  4. Assign an agent to new issues.
  5. Prototype new ideas in Spark.
  6. Branch conversations and model responses for deeper exploration.
  7. Finalize changes via pull request.

Conclusion

GitHub Copilot on the web acts as your AI-powered command and orchestration center, extending far beyond code completion in your IDE. By leveraging chat features, agent assignments, Spark prototyping, and model options, you can build a more efficient, collaborative, and AI-native developer workflow.

Ready to try it? Head to github.com/copilot and start optimizing your development experience.

This post appeared first on “The GitHub Blog”. Read the entire article here