GitHub Updates Copilot Cloud Agent to Research and Plan Before Writing Code

GitHub

Apr 6, 2026 · Updated Apr 25, 2026

GitHub updated its Copilot cloud agent to allow developers to research codebases, generate implementation plans, and commit changes directly to branches without opening a pull request. By decoupling the agent from the final pull request step, users can now iterate on complex tasks and approve technical strategies before any code is actually written.

GitHub updated Copilot cloud agent, its autonomous tool for multi-step development, with features for deep research and implementation planning. The agent is no longer restricted to pull-request workflows; it can now operate directly on a branch, allowing you to review a full diff (a representation of code changes) before deciding to submit.

This shift moves AI from a reactive autocomplete tool to a proactive collaborator. By requiring an approved implementation plan before execution, the system introduces a human-in-the-loop checkpoint. This reduces the risk of the agent pursuing incorrect technical directions during complex, repository-wide refactors or feature additions.

You can trigger these capabilities through the Agents tab in your repository or via Copilot Chat. These features are available to all paid Copilot plans, though Business and Enterprise administrators must manually enable the agent. You can now iterate on a branch until the work is ready for a formal review.

View the full update on github.blog

GitHub

@githubApr 6

GitHub Copilot cloud agent just got a lot more flexible ✨ You can now it use it to research, plan, and make code changes without needing to open a pull request first. https://t.co/zKQ4DeSiC3 https://t.co/Soi08zV4XS

19138

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

See all AI news & updates from GitHub →

Keep reading

GitHub Launches Remote Control for Copilot CLI to Steer Agents Anywhere

GitHub announced the general availability of Remote Control for Copilot CLI and Visual Studio Code sessions. This feature allows developers to monitor and approve agent actions from a phone or browser while the code runs on their local machine.

GitHub enables custom agents and reusable skills in Visual Studio

GitHub ChangelogApr 2

GitHub enables custom agents and reusable skills in Visual Studio

GitHub updated Copilot in Visual Studio with support for custom agents defined via .agent.md files and reusable agent skills. This transition turns the IDE into a programmable agentic platform where teams can standardize specialized AI behaviors and securely connect to external data through governed MCP servers.

ClaudeMar 28

Anthropic Claude Code now autonomously fixes CI failures and PR comments

Anthropic introduced a cloud-based auto-fix feature for Claude Code that monitors GitHub Pull Requests to resolve CI errors and reviewer feedback. This shifts the agent from a local tool to an autonomous background worker, allowing developers to submit code and walk away while the AI handles the iterative review cycle.

ReplitApr 18

Replit Agent Proactively Suggests and Executes Project Tasks in Parallel

Replit added a proactive task suggestion system to its AI agent that analyzes project context to recommend new features, performance fixes, and UX improvements. Users can review these plans and let the agent execute them in the background using parallel agents. This shift moves the agent from a reactive tool to a proactive collaborator that handles technical debt while the user keeps building.