Works with any MCP-compatible AI editor like Cursor, Windsurf, or Claude Desktop
GitHub’s MCP Server enables AI editors to directly interact with GitHub APIs, allowing you to automatically resolve Greptile code review comments without leaving your development environment.

​
Prerequisites

​
Setup GitHub MCP Server

​
1. Install the MCP Server

# Download the latest release
curl -L https://github.com/github/github-mcp-server/releases/latest/download/github-mcp-server-$(uname -s)-$(uname -m) -o github-mcp-server
chmod +x github-mcp-server

​
2. Configure Your Editor

Add to your Cursor settings (~/.cursor/mcp_servers.json):
{
  "mcpServers": {
    "github": {
      "command": "./github-mcp-server",
      "env": {
        "GITHUB_TOKEN": "your_github_token_here"
      }
    }
  }
}

​
Auto-Resolving Greptile Comments

​
1. Fetch PR Comments

Use your AI editor to fetch all comments from a pull request: 
@github Can you get all the comments from PR #123 in owner/repo-name? Focus on the ones from the Greptile bot.
The MCP server will retrieve all PR comments, including those from Greptile’s code review bot.

​
2. Analyze and Resolve

Ask your AI editor to resolve the comments: 
Please analyze these Greptile comments and automatically resolve them by:
1. Opening the relevant files
2. Understanding the context of each comment
3. Implementing the suggested fixes
4. Ensuring the changes don't break existing functionality

​
3. Bulk Resolution Workflow

For efficiency, you can process multiple comments at once: 
Please process all Greptile comments in this PR:
1. Group related comments by file
2. Prioritize critical issues (security, bugs) over style issues
3. Implement fixes in logical order
4. Create a summary of all changes made

​
Advanced Usage

​
Selective Resolution

Target specific types of comments: 
@github Get comments from PR #123, then resolve only the security and performance related Greptile comments, ignoring style suggestions.

​
Validation Workflow

Implement a validation step: 
After resolving each Greptile comment:
1. Run relevant tests to ensure no regressions
2. Check if the fix addresses the root cause
3. Verify the change follows project conventions
4. Create a commit message explaining the fix

​
Custom Patterns

Train your editor on your codebase patterns: 
Before resolving Greptile comments, analyze the existing codebase to understand:
- Naming conventions
- Code organization patterns  
- Testing practices
- Documentation standards

Then ensure all fixes align with these patterns.

​
Best Practices

Review Before Committing

Always review AI-generated fixes before committing, especially for complex logic changes.

Test Thoroughly

Run your test suite after resolving comments to catch any regressions.

Incremental Approach

Start with style and simple fixes before tackling complex logic issues.

Context Awareness

Ensure your AI editor understands the full context of changes, not just individual files.

​
Troubleshooting

MCP Server not responding: Verify your GitHub token has sufficient permissions and the server binary is executable. Comments not fetching: Ensure the repository name and PR number are correct, and you have read access to the repository. Fixes breaking tests: The AI may need more context about your testing patterns and validation requirements.

​
Example Workflow

# 1. Fetch PR comments
"@github List all comments from PR #456 in myorg/myrepo"

# 2. Filter for Greptile comments  
"Show me only the comments from the Greptile bot"

# 3. Resolve systematically
"Starting with the most critical issues, please resolve each Greptile comment by implementing the suggested fixes"

# 4. Validate changes
"Run the test suite and verify all fixes are working correctly"
This workflow transforms Greptile’s code review insights into actionable fixes, significantly accelerating your development cycle while maintaining code quality.