IDE-Side MCP Setup Guide¶

This guide explains how to use MCP servers directly within your IDE (VS Code, Codium, Cursor, etc.) rather than running them as containerized services.

Overview¶

Instead of running MCP servers in Docker containers as part of your development stack, you can configure your IDE to run MCP servers on-demand. This approach provides:

Direct IDE integration with GitHub, GitLab, and SonarQube
Per-project configuration flexibility
Resource efficiency (servers only run when needed)
Easier debugging and development workflow

MCP Servers from inputs/README.md¶

Based on your inputs/README.md, we'll set up these MCP servers:

GitHub MCP Server - Official GitHub integration
GitLab MCP Server - GitLab repository management
SonarQube MCP Server - Code quality analysis

Prerequisites¶

1. IDE Requirements¶

VS Code: Version 1.101+ (for MCP support)
Codium: Latest version with MCP extension support
Cursor: Latest version with MCP support
Windsurf: Latest version

2. Docker¶

All MCP servers run in Docker containers, so you need: - Docker Desktop installed and running - Access to pull public Docker images

3. Access Tokens¶

GitHub Personal Access Token¶

Go to GitHub Settings > Developer settings > Personal access tokens
Create a fine-grained personal access token with these permissions:
Repository permissions: Read access to code, issues, pull requests, actions
Account permissions: Read access to user profile
Save the token securely

GitLab Personal Access Token (Optional)¶

Go to GitLab Settings > Access Tokens
Create a token with these scopes:
read_api, read_repository, read_user
Save the token securely

SonarQube Token (Optional)¶

In your SonarQube instance, go to User > My Account > Security
Generate a new token with appropriate permissions
Save the token securely

Setup Instructions¶

Method 1: Workspace-Specific Configuration (Recommended)¶

This method configures MCP servers for a specific project/workspace.

Copy the workspace MCP configuration:

# The .vscode/mcp.json file is already created in your project
# You can customize it for your specific needs

Open your project in VS Code/Codium
Enable GitHub Copilot Agent Mode (if using VS Code with Copilot):
Open Command Palette (Ctrl+Shift+P / Cmd+Shift+P)
Type "Copilot: Toggle Agent Mode"
Enable Agent Mode
The IDE will prompt for tokens when first using MCP features

Method 2: Global User Configuration¶

This method configures MCP servers globally for all projects.

For VS Code:¶

Open VS Code Settings (Ctrl+, / Cmd+,)
Open Settings JSON (click the {} icon in the top right)
Add the MCP configuration from configs/ide-mcp/vscode-mcp.json
Save and restart VS Code

For Codium:¶

Open Codium Settings (Ctrl+, / Cmd+,)
Open Settings JSON
Add the same MCP configuration (Codium follows VS Code's format)
Save and restart Codium

Method 3: Using GitHub Copilot in Other IDEs¶

For JetBrains IDEs, Visual Studio, Eclipse, etc. with GitHub Copilot:

Add MCP configuration to your IDE's Copilot settings
Use the format from configs/ide-mcp/vscode-mcp.json (adapt syntax as needed)
Restart your IDE

Usage Examples¶

GitHub MCP Server Capabilities¶

Once configured, you can ask your AI assistant to:

# Repository Management
"Show me the recent commits in this repository"
"Create a new issue for the bug we just found"
"List all open pull requests"

# Code Analysis
"Analyze the security findings in this repo"
"Show me all Dependabot alerts"
"Check the status of GitHub Actions workflows"

# Project Management
"Create a pull request for my current branch"
"Add a comment to issue #123"
"Merge pull request #456"

GitLab MCP Server Capabilities¶

# Repository Operations
"Show me merge requests for this project"
"Create a new issue in GitLab"
"Check pipeline status"

# Project Management
"List project milestones"
"Show recent activity"
"Create a merge request"

SonarQube MCP Server Capabilities¶

# Code Quality
"Show code quality metrics for this project"
"List all security vulnerabilities"
"Analyze code coverage reports"

# Issue Management
"Show critical issues in the codebase"
"Generate a quality gate report"
"List technical debt items"

Configuration Options¶

GitHub Server Options¶

You can customize the GitHub MCP server by modifying the environment variables:

{
  "env": {
    "GITHUB_PERSONAL_ACCESS_TOKEN": "${input:github_token}",
    "GITHUB_TOOLSETS": "repos,issues,pull_requests,actions,code_security,context",
    "GITHUB_READ_ONLY": "false",
    "GITHUB_HOST": "https://github.com"  // For GitHub Enterprise
  }
}

Available toolsets: - context - User context and GitHub info (recommended) - repos - Repository operations
- issues - Issue management - pull_requests - PR management - actions - GitHub Actions/CI-CD - code_security - Security scanning - discussions - GitHub Discussions - notifications - Notification management - orgs - Organization management - users - User operations

GitLab Server Options¶

{
  "env": {
    "GITLAB_TOKEN": "${input:gitlab_token}",
    "GITLAB_URL": "https://gitlab.com",  // Change for self-hosted
    "GITLAB_API_VERSION": "v4"
  }
}

SonarQube Server Options¶

{
  "env": {
    "SONARQUBE_TOKEN": "${input:sonarqube_token}", 
    "SONARQUBE_URL": "http://localhost:9000",  // Your SonarQube instance
    "SONARQUBE_ORGANIZATION": "your-org"       // For SonarCloud
  }
}

Troubleshooting¶

Common Issues¶

"Docker image not found"
Ensure Docker is running
Pull images manually: docker pull ghcr.io/github/github-mcp-server
"Authentication failed"
Verify your access tokens are correct and not expired
Check token permissions match requirements
"MCP server not responding"
Check Docker logs: docker logs <container-id>
Verify network connectivity to GitHub/GitLab/SonarQube
"VS Code not recognizing MCP configuration"
Ensure VS Code version is 1.101+
Restart VS Code after configuration changes
Check the JSON syntax is valid

Debug Mode¶

Enable debug logging by adding to your MCP server configuration:

{
  "env": {
    "DEBUG": "true",
    "LOG_LEVEL": "debug"
  }
}

Testing MCP Connection¶

You can test MCP servers directly:

# Test GitHub MCP server
docker run -it --rm -e GITHUB_PERSONAL_ACCESS_TOKEN=your_token ghcr.io/github/github-mcp-server

# Test GitLab MCP server  
docker run -it --rm -e GITLAB_TOKEN=your_token zereight/gitlab-mcp

# Test SonarQube MCP server
docker run -it --rm -e SONARQUBE_TOKEN=your_token sonarsource/sonarqube-mcp-server

Security Considerations¶

Token Storage: IDE configurations store tokens securely, but avoid committing them to version control
Workspace Isolation: Use workspace-specific configurations for sensitive projects
Read-Only Mode: Enable read-only mode for GitHub server in production environments
Network Access: MCP servers run in containers with network access to external services

Alternative: Remote MCP Servers¶

For GitHub, you can also use the hosted remote MCP server:

{
  "servers": {
    "github": {
      "type": "http",
      "url": "https://api.githubcopilot.com/mcp/"
    }
  }
}

This doesn't require running Docker containers locally but requires GitHub Copilot subscription.

Next Steps¶

Choose your setup method (workspace-specific recommended)
Configure your IDE with the appropriate MCP settings
Obtain required access tokens for the services you want to use
Test the integration by asking your AI assistant to interact with GitHub/GitLab/SonarQube
Explore advanced features like automated issue creation, PR management, and code quality analysis

The IDE-side MCP integration provides powerful capabilities for AI-assisted development workflows while keeping the setup flexible and resource-efficient.