How We Orchestrate with AI

This guide covers how we use Claude Code connected to GitHub MCP (Model Context Protocol) to manage our development workflow - PRs, issues, and code reviews - with AI assistance.

Core idea: AI can handle the mechanical parts of development workflows, letting you focus on the creative and strategic work.

What is AI Orchestration?

AI orchestration means using AI tools to automate and assist with development workflows:

Developer Intent → Claude Code → GitHub MCP → GitHub Actions
      ↓                ↓              ↓
   "Fix this bug"   Understands    Creates PR
                    context        Updates issues
                                   Manages workflow

What is MCP?

Model Context Protocol (MCP) is a standard for connecting AI assistants to external tools and services. The GitHub MCP server gives Claude Code access to:

• Repository information
• Issues and pull requests
• File contents
• Commit history
• GitHub Actions

Setting Up Claude Code with GitHub MCP

Prerequisites

1. Claude Code installed
2. GitHub CLI authenticated (gh auth login)
3. MCP server configured

Configuration

Add the GitHub MCP server to your Claude Code configuration:

{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "<your-token>"
      }
    }
  }
}

Common Workflows

Creating a PR with Context

Instead of manually writing PR descriptions:

"Create a PR for my current changes. Include what changed and why."

Claude Code will:

1. Read your changes
2. Understand the context
3. Generate a descriptive PR
4. Link related issues

Issue Triage

"Look at the open issues and categorize them by priority and type."

Claude Code will:

1. Read all open issues
2. Analyze content and labels
3. Suggest categorization
4. Optionally apply labels

Code Review Assistance

"Review PR #123 and summarize the changes."

Claude Code will:

1. Fetch the PR diff
2. Analyze the changes
3. Identify potential issues
4. Summarize for human review

Decision Framework

When to Use AI Orchestration

When to Use Manual Workflow

• Security-sensitive changes
• Breaking changes that need careful communication
• When you need full control over the narrative
• Learning (doing it manually builds understanding)

Best Practices

1. Review AI-Generated Content

Always review before submitting:

• PR descriptions may miss context you have
• Issue labels may not match team conventions
• Comments should reflect your voice

2. Start Small

Begin with low-stakes tasks:

• Draft PR descriptions (edit before submitting)
• Summarize changes for your own understanding
• Generate commit messages

3. Provide Context

The more context you give, the better the output:

• Reference related issues
• Explain the "why" not just the "what"
• Mention team conventions

Limitations

Quick Reference

Resources

• Model Context Protocol
• GitHub MCP Server
• Claude Code Documentation