GitHub
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
作者による説明(英語の原文)
GitLab CI/CD MCP — pipelines, jobs, schedules, MRs, files. Any GitLab (SaaS or self-hosted).
この3つを1つのスコアにまとめることはしません。意見、測定、人気はそれぞれ別のものだからです。
このサーバーのやさしい解説はまだ書いていません。以下はレジストリから読み取った事実です。
使っているAIアプリを選んで、設定をコピーしてください。
GITLAB_URL, GITLAB_TOKEN, GITLAB_PROJECT_PATH, GITLAB_SSL_VERIFY「秘密」と表示された値は、あなたのアカウントの鍵です。絶対に人に教えたり、チャットに貼り付けたりしないでください。
必要なもの
どのアプリを使っていますか?
上でアプリを選んでください。選んだアプリを覚えておくので、次からはボタンに「…にインストール」と表示されます。
Claude Desktop → Settings → Developer → Edit Config を開き、claude_desktop_config.json にこれを貼り付けます
{
"mcpServers": {
"gitlab-ci": {
"command": "uvx",
"args": [
"gitlab-ci-mcp"
],
"env": {
"GITLAB_URL": "your-value",
"GITLAB_TOKEN": "your-value",
"GITLAB_PROJECT_PATH": "your-value",
"GITLAB_SSL_VERIFY": "your-value"
}
}
}
}ターミナルでこれを実行します
claude mcp add gitlab-ci --env GITLAB_URL=your-value --env GITLAB_TOKEN=your-value --env GITLAB_PROJECT_PATH=your-value --env GITLAB_SSL_VERIFY=your-value -- uvx gitlab-ci-mcp動かす前に、空欄の値を入力してください。
プロジェクトの .cursor/mcp.json、または ~/.cursor/mcp.json に貼り付けます
{
"mcpServers": {
"gitlab-ci": {
"command": "uvx",
"args": [
"gitlab-ci-mcp"
],
"env": {
"GITLAB_URL": "your-value",
"GITLAB_TOKEN": "your-value",
"GITLAB_PROJECT_PATH": "your-value",
"GITLAB_SSL_VERIFY": "your-value"
}
}
}
}ブラウザがアプリを開いてよいか確認し、そのあとアプリが設定を表示して確認を求めます。キーの値は空欄のままなので、自分で入力してください。
プロジェクトの .vscode/mcp.json に貼り付けます
{
"servers": {
"gitlab-ci": {
"type": "stdio",
"command": "uvx",
"args": [
"gitlab-ci-mcp"
],
"env": {
"GITLAB_URL": "your-value",
"GITLAB_TOKEN": "your-value",
"GITLAB_PROJECT_PATH": "your-value",
"GITLAB_SSL_VERIFY": "your-value"
}
}
}
}ブラウザがアプリを開いてよいか確認し、そのあとアプリが設定を表示して確認を求めます。キーの値は空欄のままなので、自分で入力してください。
これはパソコンで動くサーバーですが、ChatGPTがつなげるのはインターネット上のサーバーだけです。ほかのアプリを選んでください。
選んだアプリは、この端末のCookieに保存されます。アカウントは不要です。
GITLAB_PROJECT_PATH envGITLAB_SSL_VERIFY envこの設定は、作者のレジストリ登録情報から作成しています。READMEと違う場合は、READMEに従ってください。
パソコンで動くサーバーです。インストールすると、ツール一覧がAIアプリに表示されます。
誰でも見られる情報をチェックしており、コードは読んでいません。当てはまらない項目は数に入れていません。 · 2026年11月6日にチェック
レジストリ名は、リポジトリを所有するGitHubアカウント mshegolev で確認済みです
このリポジトリはまだGitHubから読み取っていません
このリポジトリはまだGitHubから読み取っていません
これはパソコンで動くため、インターネット上のエンドポイントはありません
パソコンで動き、代わりにローカルの設定を使います
パソコンで動くサーバーです。ツール一覧を読み取れるのは、インターネット上のエンドポイントがある場合だけです
レジストリでこのカテゴリにある有効なすべてのサーバーと比較しています。スターは、公開者がリポジトリの所有者である場合だけ数えます。
当てはまる基本チェック1項目のうち1項目に合格しました。コードは読んでいません。キーやトークンを追加する前に、サーバーがどんなアクセスを求めているかを確認し、必要な権限だけを持つキーを使ってください。
入力が必要なもの:GITLAB_URL, GITLAB_TOKEN
あなたのパソコンで動きます。必要なときにAIアプリが起動し、データは作者のサーバーを通りません(サーバー自体がオンラインサービスを呼び出す場合を除きます)。
GitHubリポジトリの最終更新は–、最新バージョン(0.5.1)がレジストリに届いたのは7 か月前です。
Claude Desktop、Claude Code、Cursor、VS CodeなどのMCP対応アプリで使えます。パソコンにPython (uv)も必要です。
同じカテゴリには:GitHub, Desktop Commander, Cloudflare MCP
同じカテゴリで、ほかの作者が作ったサーバー。スターの多い順。
カードの見方:★ GitHubスターi3つの柱:レビュー · チェック · 利用状況i更新=最新バージョンをレジストリに公開した日i
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
MCP server for terminal commands, file operations, and process management
Cloudflare MCP servers
作者のリポジトリから書かれたとおりにコピーしたものです。PopNx MCPは内容を確認・推奨していません。HTML、画像、スクリプトは取り除いています。リンクはこのサイトの外へ移動します。
MCP server for GitLab CI/CD. Lets an LLM agent (Claude Code, Cursor, OpenCode, DevX Agent, etc.) work with pipelines, jobs, schedules, branches, tags, merge requests and repository files.
Python and FastMCP. Local stdio remains the default; opt-in Streamable HTTP supports shared deployments with one GitLab PAT per caller.
Works with any GitLab — SaaS gitlab.com or self-hosted / on-prem. Designed with corporate networks in mind: configurable NO_PROXY handling, optional SSL-verify toggle, per-project scoping via env vars.
readOnlyHint / destructiveHint / idempotentHint / openWorldHint so MCP clients can classify operations (e.g. ask for confirmation only on destructive ones like gitlab_merge_mr, gitlab_delete_schedule).outputSchema and every result carries structuredContent alongside a pre-rendered markdown text block. Clients that can render structured data use it; agents that prefer compact text get the markdown. No response_format parameter needed.ToolError messages (e.g. _"GitLab authentication failed… verify GITLAB_TOKEN has api scope"_) and surfaced as isError=True results.project_path by default. Set GITLAB_ALLOW_DEFAULT_PROJECT=true to opt into the legacy environment fallback.pagination block with page, total, has_more, next_page and a next-page hint in the markdown footer.X-GitLab-Token; GitLab API calls and cached clients are isolated by PAT. Stdio remains single-user with GITLAB_TOKEN.gitlab://project/info and gitlab://project/ci-config mirror common lookups for clients that prefer the Resource model over tools.python-gitlab HTTP sessions are closed cleanly on server shutdown via an asynccontextmanager lifespan hook.gitlab_get_job_log accepts grep_pattern + grep_context (surrounding lines) for regex-filtering megabyte-scale CI logs without pulling the whole trace into agent context.FastMCP automatically runs synchronous tools in a worker thread (anyio.to_thread.run_sync), so they do not block the asyncio event loop — python-gitlab is a synchronous library and wrapping every call in asyncio.to_thread ourselves would be ceremony. Tools that benefit from the MCP Context (progress, info logs) are written as async def and explicitly wrap python-gitlab calls with asyncio.to_thread.
24 tools covering the everyday CI/CD surface:
Pipelines gitlab_list_pipelines · gitlab_get_pipeline · gitlab_get_pipeline_jobs · gitlab_get_job_log · gitlab_trigger_pipeline · gitlab_retry_pipeline · gitlab_cancel_pipeline · gitlab_pipeline_health
Schedules gitlab_list_schedules · gitlab_create_schedule · gitlab_update_schedule · gitlab_delete_schedule
Branches & tags gitlab_list_branches · gitlab_list_tags · gitlab_compare_branches
Merge requests gitlab_list_merge_requests · gitlab_get_merge_request · gitlab_get_merge_request_changes · gitlab_create_merge_request · gitlab_merge_mr
Repository & project gitlab_get_file · gitlab_list_repository_tree · gitlab_project_info
Package gitlab_ci_mcp_check_updates
gitlab_pipeline_health returns a ready-to-read summary over 7/30 days:
Last 7d: 96.4% up | 27/28 success
Last 30d: 92.1% | 105/114 success Last 10: success success success failed success ...
Handy for on-call / triage: покажи health master за последние 7 дней.
Requires Python 3.10+.
# via uvx (recommended)
uvx --from gitlab-ci-mcp gitlab-ci-mcp
pipx install gitlab-ci-mcp
All config is via environment variables:
| Variable | Required | Description |
|---|---|---|
GITLAB_URL | yes | Base URL, e.g. https://gitlab.example.com |
GITLAB_TOKEN | stdio only | Personal Access Token with api scope. In HTTP mode, each client sends its own X-GitLab-Token header. |
MCP_TRANSPORT | no | stdio (default) or streamable-http |
MCP_HTTP_HOST | no | HTTP bind host; defaults to 127.0.0.1 |
MCP_HTTP_PORT | no | HTTP bind port; defaults to 8000 |
GITLAB_PROJECT_PATH | no | Legacy fallback project; used only with GITLAB_ALLOW_DEFAULT_PROJECT=true |
GITLAB_ALLOW_DEFAULT_PROJECT | no | Set to true to opt into the legacy GITLAB_PROJECT_PATH fallback |
MCP_GITLAB_UPDATE_CHECK_ENABLED | no | false by default; set to true to allow the explicit update-check tool to query PyPI |
GITLAB_SSL_VERIFY | no | true (default) / false |
GITLAB_NO_PROXY_DOMAINS | no | Comma-separated domains to add to NO_PROXY (useful in corp networks behind a local HTTP proxy — e.g. .corp.example.com,gitlab.internal) |
Every tool accepts a project_path arg. It is required by default so a call cannot silently target an environment-wide project. For legacy deployments, set GITLAB_ALLOW_DEFAULT_PROJECT=true together with GITLAB_PROJECT_PATH to opt into the old fallback behavior.
Update checking is disabled by default. Set MCP_GITLAB_UPDATE_CHECK_ENABLED=true to enable the explicit gitlab_ci_mcp_check_updates tool. When enabled, it makes one bounded request to fixed PyPI JSON, caches a result for five minutes, reports failures and unsupported versions, and never installs anything. No startup or other tool call performs this request. If called while disabled, the tool reports disabled without making a network request.
Pipeline health success rate uses completed pipelines only: success and failed are in the denominator; running, pending, manual, canceled, skipped, and other non-completed statuses are excluded. A missing comparison period or a sample without completed pipelines is reported as insufficient data, not as a zero rate or a flat trend.
For a shared server, run with MCP_TRANSPORT=streamable-http. Each caller must send their own GitLab PAT in X-GitLab-Token; the server never falls back to the process GITLAB_TOKEN in this mode. GitLab applies that PAT's identity and permissions to each API operation. The header is a custom credential mechanism, not MCP OAuth; clients need support for configured custom headers.
The server binds to loopback (127.0.0.1:8000) by default. Expose it through an HTTPS reverse proxy, restrict network access, and redact the PAT header from proxy, tracing, and error logs. Initialize and tool discovery only check that the header is non-empty; GitLab validates the PAT when an API call is made.
Example for Claude Code (provide GITLAB_PAT through a secret manager; do not put a literal token in shell history):
claude mcp add --transport http \
--header "X-GitLab-Token: ${GITLAB_PAT}" \ gitlab https://mcp.example.com/mcp
The client must attach the header to every request. Each GitLab tool still requires an explicit project_path. Fixed resources gitlab://project/info and gitlab://project/ci-config have no project argument; they require the explicit compatibility setting GITLAB_ALLOW_DEFAULT_PROJECT=true and GITLAB_PROJECT_PATH.
Full credential details and limitations: docs/per-user-auth.md.
Full walkthrough: docs/claude-code.md — prerequisites, two install paths, multi-project setup, self-hosted GitLab behind a corp proxy, troubleshooting, uninstall.
Short version:
claude mcp add --transport stdio \
--env GITLAB_URL=https://gitlab.example.com \ --env GITLAB_TOKEN="$GITLAB_PAT" \ gitlab -- uvx --from gitlab-ci-mcp gitlab-ci-mcp
Each GitLab tool call must include project_path, for example project_path="my-org/my-repo". For legacy default-project behavior, also set GITLAB_PROJECT_PATH and GITLAB_ALLOW_DEFAULT_PROJECT=true.
Or in ~/.claude.json / project .mcp.json:
{
"mcpServers": { "gitlab": { "type": "stdio", "command": "uvx", "args": ["--from", "gitlab-ci-mcp", "gitlab-ci-mcp"], "env": { "GITLAB_URL": "https://gitlab.example.com", "GITLAB_TOKEN": "${GITLAB_PAT}", "GITLAB_PROJECT_PATH": "my-org/my-repo", "GITLAB_ALLOW_DEFAULT_PROJECT": "true", "GITLAB_SSL_VERIFY": "true" } } } }
Check:
claude mcp list
Same idea — point the MCP config to uvx --from gitlab-ci-mcp gitlab-ci-mcp with the env vars above. See each tool's own MCP config syntax.
что сломалось в последнем pipeline master
покажи health master за 7 дней для проекта my-org/other-repo
создай MR из feature/foo в master с title "feat: foo"
покажи содержимое .gitlab-ci.yml из master
GitLab enforces a per-user rate limit (typically 2000 req/h for the REST API, configurable by the admin — see your instance's /admin/application_settings/network).
python-gitlab HTTP session per GitLab URL, project, and PAT fingerprint. Different users never reuse one another's cached client.per_page=20 to keep a single call within a small number of API requests.429 Too Many Requests, the error handler returns an actionable message — wait and try again with larger per_page or fewer calls.When your laptop has a local HTTP proxy (e.g. http://127.0.0.1:3128 for corp web access) but GitLab is on the intranet, the proxy intercepts and kills internal requests. Two options:
GITLAB_NO_PROXY_DOMAINS — the server will add them to NO_PROXY at startup and clear `HTTP_PROXY`/`HTTPS_PROXY` from its own process so they don't affect GitLab traffic.HTTP_PROXY="" etc. in the MCP env section.git clone https://github.com/mshegolev/gitlab-ci-mcp
cd gitlab-ci-mcp python -m venv .venv && . .venv/bin/activate pip install -e '.[dev]' pytest
Run the server directly (stdio transport is the default and waits on stdin for MCP messages):
GITLAB_URL=... GITLAB_TOKEN=... gitlab-ci-mcp
MIT — see LICENSE.
Built on python-gitlab and the MCP Python SDK.