Skip to main content

Model Context Protocol (MCP)

MCPs let your AI tool access your codebase, integrate with external services, and perform advanced tasks. This guide shows what MCPs are available and how to configure them for your tool.

What is MCP?

Model Context Protocol allows your AI tool to:

  • Search your code and documentation
  • Access git history and repository information
  • Browse and read files
  • Interact with external services (GitLab, Jira, AWS)
  • Run custom tools and scripts

Recomended MCPs

MCPWhat It DoesRequires
GitLabCreate/manage merge requests, issues, search codeGitLab account & token
AtlassianRead/interact with Jira and ConfluenceAtlassian account & token
FigmaAccess Figma files, designs, and prototypesFigma account & token
AWS DocsSearch and read AWS documentationInternet access
WizSecurity and cloud risk analysisWiz account & token
TrivySecurity scanning for vulnerabilitiesTrivy installed + MCP plugin
Chrome DevToolsInteract with Chrome browser via DevTools ProtocolChrome browser + npx
PlaywrightBrowser automation and testingnpx
Fetch MCP ServerFetch and read web contentuv/uvx installed

Setup by Tool

Configure MCPs for your AI tool by selecting your tool in each section below.

GitLab MCP

Create and manage merge requests, issues, search code, and access GitLab projects.

Setup: First install the glab CLI tool and authenticate:

# Install glab (macOS)
brew install glab

# Or install manually
# Visit https://github.com/profclems/glab/releases

# Authenticate with GitLab
glab auth login

# Follow the prompts to authenticate with your GitLab account
~/.claude/config.json
{
"mcp_servers": {
"gitlab": {
"command": "glab",
"args": ["mcp", "serve"]
}
}
}

Atlassian MCP

Read and interact with Jira and Confluence.

~/.claude/config.json
{
"mcp_servers": {
"atlassian": {
"type": "http",
"url": "https://mcp.atlassian.com/v1/mcp"
}
}
}

Figma MCP

Access Figma files, designs, and prototypes.

~/.claude/config.json
{
"mcp_servers": {
"figma": {
"url": "https://mcp.figma.com/mcp"
}
}
}

AWS Docs MCP

Search and read AWS documentation using the official AWS Documentation MCP server.

Setup: Install uv (required dependency):

# Install uv (universal Python environment manager)
curl -LsSf https://astral.sh/uv/install.sh | sh
~/.claude/config.json
{
"mcp_servers": {
"aws-docs": {
"command": "uvx",
"args": ["awslabs.aws-documentation-mcp-server@latest"],
"env": {
"FASTMCP_LOG_LEVEL": "ERROR",
"AWS_DOCUMENTATION_PARTITION": "aws",
"MCP_USER_AGENT": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36"
}
}
}
}

Environment Variables:

  • FASTMCP_LOG_LEVEL: Set to ERROR for cleaner output (options: DEBUG, INFO, WARNING, ERROR, CRITICAL)
  • AWS_DOCUMENTATION_PARTITION: Set to aws for global docs or aws-cn for China region docs
  • MCP_USER_AGENT: Custom User-Agent string (useful for corporate networks with proxy restrictions)

Wiz MCP

Security and cloud risk analysis from Wiz.

~/.claude/config.json
{
"mcp_servers": {
"wiz": {
"type": "http",
"url": "https://mcp.app.wiz.io"
}
}
}

Trivy MCP

Security scanning for vulnerabilities and misconfigurations.

Setup: First install Trivy and the MCP plugin:

# Install Trivy (if not already installed)
brew install trivy # macOS
# or visit https://github.com/aquasecurity/trivy

# Install MCP plugin
trivy plugin install mcp
~/.claude/config.json
{
"mcp_servers": {
"trivy": {
"command": "trivy",
"args": ["mcp"]
}
}
}

Chrome DevTools MCP

Interact with Chrome browser via DevTools Protocol for browser automation and testing.

Setup: Requires Node.js and npx (comes with Node.js):

# Verify npx is available
npx --version
~/.claude/config.json
{
"mcp_servers": {
"chrome-devtools": {
"command": "npx",
"args": ["chrome-devtools-mcp@latest"]
}
}
}

Playwright MCP

Browser automation and testing with Playwright.

Setup: Requires Node.js and npx (comes with Node.js):

# Verify npx is available
npx --version
~/.claude/config.json
{
"mcp_servers": {
"playwright": {
"command": "npx",
"args": ["@playwright/mcp@latest"]
}
}
}

Fetch MCP Server

Fetch and read web content from URLs.

Setup: Install uv or uvx (required for running the MCP server):

# Install uv (universal Python environment manager)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Or install uvx (comes with uv)
# uvx allows running Python tools without installation
~/.claude/config.json
{
"mcp_servers": {
"fetch": {
"command": "uvx",
"args": ["mcp-server-fetch"]
}
}
}