GitLab MCP Server
Enables AI assistants to interact with GitLab for managing projects, branches, issues, and merge requests. It provides tools for searching code and performing file operations like reading and writing directly within repositories.
README
GitLab MCP Server
A Model Context Protocol (MCP) server for GitLab integration with Claude Code and AI assistants.
Features
- Projects: List and get project details
- Branches: List and create branches
- Issues: Create and list issues
- Merge Requests: Create and list MRs
- Files: Read and write repository files
- Search: Search code in repositories
Requirements
Installation
The server is published on npm and can be run directly:
# Using Bun (recommended)
bunx @carmeloricarte/gitlab-mcp-server
# Using Node.js
npx @carmeloricarte/gitlab-mcp-server
No need to clone the repository or install dependencies manually.
Configuration
Set the following environment variables:
| Variable | Required | Default | Description |
|---|---|---|---|
GITLAB_TOKEN |
Yes | - | GitLab Personal Access Token |
GITLAB_HOST |
No | https://gitlab.com |
GitLab instance URL |
For self-hosted GitLab with self-signed certificates:
export NODE_TLS_REJECT_UNAUTHORIZED=0
Environment Variables Setup
There are two approaches to configure credentials:
Option A: Variables in MCP Config (Simplest)
Pass all variables directly in the MCP configuration. Easiest for quick setup on new machines.
⚠️ Note: Token is stored in the config file (local, private). Acceptable for personal use.
Option B: System Environment Variables (More Secure)
Keep sensitive tokens at system/user level, only pass non-sensitive values in MCP config.
<details> <summary><strong>Windows (PowerShell)</strong></summary>
# Set permanently for current user
[Environment]::SetEnvironmentVariable("GITLAB_TOKEN", "glpat-your-token", "User")
[Environment]::SetEnvironmentVariable("GITLAB_HOST", "https://your-gitlab.com", "User")
# Verify
[Environment]::GetEnvironmentVariable("GITLAB_TOKEN", "User")
Restart your terminal/IDE after setting variables.
</details>
<details> <summary><strong>macOS / Linux</strong></summary>
# Add to ~/.zshrc (macOS) or ~/.bashrc (Linux)
echo 'export GITLAB_TOKEN="glpat-your-token"' >> ~/.zshrc
echo 'export GITLAB_HOST="https://your-gitlab.com"' >> ~/.zshrc
# Reload
source ~/.zshrc
# Verify
echo $GITLAB_TOKEN
</details>
IDE / Tool Configuration
💡 Tip: Use
bunxif you have Bun installed, ornpxfor Node.js. Both work identically.
⚠️ Important:
${VARIABLE}syntax does NOT work in most MCP configs - values are treated as literal strings, not resolved. Use Option A (hardcoded values) or Option B (system variables that the server reads fromprocess.env).
⚠️ Windows Configuration
On Windows, you must wrap npx or bunx commands with cmd /c. Use this format:
{
"command": "cmd",
"args": ["/c", "npx", "-y", "@carmeloricarte/gitlab-mcp-server"]
}
Or with Bun:
{
"command": "cmd",
"args": ["/c", "bunx", "@carmeloricarte/gitlab-mcp-server"]
}
Important: The package name must always be the last argument in the args array.
Claude Code CLI
<details> <summary><strong>Option A: All variables in config</strong></summary>
claude mcp add-json GitLab '{
"type": "stdio",
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}'
</details>
<details> <summary><strong>Option B: Token from system env</strong></summary>
Set GITLAB_TOKEN as system variable (see above), then:
claude mcp add-json GitLab '{
"type": "stdio",
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}'
</details>
Verify:
claude mcp list
# Expected: GitLab: bunx ... - ✓ Connected
VS Code (with MCP extension)
Edit ~/.vscode/mcp.json or .vscode/mcp.json in your project:
{
"servers": {
"GitLab": {
"type": "stdio",
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}
}
}
Cursor
Edit ~/.cursor/mcp.json:
{
"mcpServers": {
"GitLab": {
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}
}
}
Zed
Edit ~/.config/zed/settings.json (macOS/Linux) or %APPDATA%\Zed\settings.json (Windows):
{
"context_servers": {
"GitLab": {
"command": {
"path": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}
}
}
}
OpenCode
Edit ~/.config/opencode/config.json:
{
"mcp": {
"servers": {
"GitLab": {
"type": "stdio",
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}
}
}
}
Codex (OpenAI CLI)
Edit ~/.codex/config.json:
{
"mcpServers": {
"GitLab": {
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}
}
}
Windsurf
Edit ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"GitLab": {
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}
}
}
Claude Desktop
Edit the Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"GitLab": {
"command": "bunx",
"args": ["@carmeloricarte/gitlab-mcp-server"],
"env": {
"GITLAB_HOST": "https://your-gitlab.com",
"GITLAB_TOKEN": "glpat-your-token",
"NODE_TLS_REJECT_UNAUTHORIZED": "0"
}
}
}
}
Available Tools
Projects
list_projects- List accessible GitLab projectsget_project- Get details of a specific project
Branches
list_branches- List branches in a projectcreate_branch- Create a new branch
Issues
create_issue- Create a new issuelist_issues- List issues in a project
Merge Requests
create_merge_request- Create a merge requestlist_merge_requests- List merge requests
Files
get_file- Get file contents from repositorycreate_or_update_file- Create or update a file
Search
search_code- Search for code in a project
Development
If you want to contribute or run the server locally for development:
Clone and Install
git clone https://github.com/CarmeloRicarte/gitlab-mcp-server.git
cd gitlab-mcp-server
bun install
Project Structure
src/
├── index.ts # Entry point
├── server.ts # MCP server setup
├── config.ts # Environment configuration
├── client/
│ └── gitlab-client.ts # GitLab API client
├── tools/
│ ├── index.ts # Tool registration
│ ├── projects.ts
│ ├── branches.ts
│ ├── issues.ts
│ ├── merge-requests.ts
│ ├── files.ts
│ └── search.ts
└── types/
└── gitlab.ts # TypeScript types
scripts/
└── add-shebang.js # Adds Node shebang to compiled output
tests/
├── setup.ts # Test utilities & mocks
├── client/
│ └── gitlab-client.test.ts
└── tools/
└── *.test.ts
Run Server
bun run start
# or with hot reload
bun run dev
Run Tests
# Run all tests
bun test
# Run with watch mode
bun test:watch
# Run with coverage
bun test:coverage
Type Check
bun run typecheck
Build
# Build for production (compiles to dist/index.js with Node.js compatibility)
bun run build
# Add shebang to compiled output (done automatically on publish)
bun run add-shebang
Publish
Publishing to npm is automated via prepublishOnly:
npm version patch # or minor/major
npm publish
This automatically:
- Compiles TypeScript to JavaScript (
bun run build) - Adds
#!/usr/bin/env nodeshebang (bun run add-shebang)
Architecture
The server uses dependency injection for the GitLab client, making it easy to mock in tests:
import { GitLabClient } from "./src/client/gitlab-client";
import { createServer } from "./src/server";
// For testing with a mock client
const mockClient = new GitLabClient({
apiBase: "https://mock.gitlab.com/api/v4",
token: "test-token",
});
const server = createServer(mockClient);
Contributing
Contributions are welcome! Feel free to open issues and pull requests.
- Fork the repository
- Create your feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'feat: add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
Please make sure to:
- Follow the existing code style
- Add tests for new features
- Update documentation as needed
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。