Particle.news

Download on the App Store

GitHub Unveils AI-Powered Copilot Agent for Autonomous Coding Tasks

The new Copilot Agent, available in preview for Pro+ and Enterprise users, automates bug fixes, feature additions, and more, leveraging Anthropic’s Claude 3.7 Sonnet model with developer oversight.

Image

Overview

  • GitHub's Copilot Agent can autonomously perform low-to-medium complexity coding tasks, including bug fixes, feature additions, and documentation improvements.
  • The agent automates workflows by setting up virtual environments, analyzing codebases, and providing detailed session logs for transparency.
  • It integrates with pull request and issue discussions, following custom repository instructions to align with project standards and intent.
  • Anthropic’s Claude 3.7 Sonnet AI model powers the agent, which is available in preview to Copilot Pro+ and Enterprise subscribers via GitHub’s platform and CLI tools.
  • Developers retain control through review processes, with the agent addressing feedback iteratively, and GitHub has open-sourced its VS Code extensions for further customization.