Work on GitHub issues

Rover integrates with GitHub to create tasks directly from issues. This allows you to assign GitHub issues to local AI coding agents and maintain a clear link between the issue and the task result.

Prerequisites

To create Rover tasks from GitHub issues, you need either:

• GitHub CLI (gh): Recommended for private repositories and GitHub Enterprise. Install from cli.github.com.
• Public repository access: For public repositories, Rover can fetch issues directly from the GitHub API without authentication.

Create a task from an issue

Use the --from-github flag to create a task using the content of a GitHub issue as the task description.

Creating a task from GitHub

1. 1

   Navigate to your project directory

   cd ~/my-project

2. 2

   Create a task from a GitHub issue by specifying the issue number

   rover task --from-github 42

   Rover fetches the issue title and body from GitHub and uses them as the task description.

3. 3

   Follow the task progress

   rover logs -f 1

4. 4

   Once the task completes, merge the changes into your branch or push them to a remote branch

   rover merge 1
   # or
   rover push 1

   See Merge changes from tasks for detailed guidance on reviewing and integrating task results.

Tip

You can combine --from-github with other task options like --agent or --source-branch:

rover task --from-github 42 --agent claude:opus --source-branch develop

Understanding GitHub metadata

When you create a task from a GitHub issue, Rover stores metadata about the source issue. This metadata includes:

Field	Description
type	Source type, always github for GitHub issues
id	The issue number
url	Direct link to the GitHub issue
ref	Reference data including owner, repo, and number

This metadata maintains the connection between your task and its originating issue throughout the development cycle.

Inspect GitHub metadata

Use the rover inspect command to view the GitHub issue linked to a task.

Viewing GitHub metadata

1. 1

   Inspect a task to see its details, including the linked GitHub issue

   rover inspect 1

   The output includes a “GitHub Issue” field with a direct link to the original issue:

   Details
   -------
   · ID: 1 (a1b2c3d4-5678-90ab-cdef-1234567890ab)
   · Title: Implement user authentication
   · Status: Completed
   · Workflow: swe
   · Created At: 1/27/2026, 10:30:00 AM
   · GitHub Issue: https://github.com/acme/webapp/issues/42
   · Completed At: 1/27/2026, 11:15:00 AM
   
   Workspace
   ---------
   · Branch Name: rover/task-1-xK9mPq2wLz4N
   · Git Workspace path: ~/.rover/data/projects/acme-webapp-8f3a1b2c/workspaces/1

2. 2

   For programmatic access, use the --json flag to get the full source metadata

   rover inspect 1 --json

   The JSON output includes the complete source object:

   {
     "success": true,
     "id": 1,
     "title": "Implement user authentication",
     "status": "COMPLETED",
     "branchName": "rover/task-1-xK9mPq2wLz4N",
     "source": {
       "type": "github",
       "id": "42",
       "url": "https://github.com/acme/webapp/issues/42",
       "ref": {
         "owner": "acme",
         "repo": "webapp",
         "number": 42
       }
     }
   }

Working with custom workflows

When using a custom workflow that requires additional inputs beyond a description, Rover extracts the required information from the GitHub issue body. The AI agent analyzes the issue content and maps it to the workflow inputs.

For example, if your workflow requires description and acceptance_criteria inputs, include both in your GitHub issue body:

## Description
Add a login form with email and password fields.

## Acceptance Criteria
- Email validation
- Password minimum 8 characters
- Show error messages on invalid input

Rover parses the issue and populates the workflow inputs accordingly.

Best practices

To get the best results when creating tasks from GitHub issues:

• Write detailed issue descriptions: Include context, requirements, and acceptance criteria
• Use clear formatting: Markdown headings and bullet points help the agent understand the structure
• Reference specific files: Mention file paths or function names when relevant
• Include examples: Provide code snippets or expected output when applicable
• Keep scope focused: Smaller, well-defined issues lead to better implementations

Relevant Concepts

Task Learn about Rover tasks and how they work

Workflow Discover how workflows guide coding agents

Related Guides

Implement a new feature Create and assign tasks to AI coding agents

Iterate on tasks Refine and improve task implementations

Merge changes from tasks Integrate completed tasks into your codebase