Jenkins MCP Server
Provides Jenkins CI/CD integration for AI assistants through the Model Context Protocol, enabling job management, build control, and system administration via natural language commands.
README
Jenkins MCP Server
A Model Context Protocol (MCP) server that provides Jenkins CI/CD integration for Cursor IDE and other MCP-compatible AI assistants.
Features
- 27 MCP tools covering jobs, builds, pipelines, nodes, credentials, views, system info, and more
- 3 access modes --
read-only,standard,admin-- so you can share safely with any audience - Job creation from templates -- create Freestyle, Pipeline, Pipeline-SCM, and Multibranch jobs without writing XML
- CLI tool for quick Jenkins queries from the terminal
- Python 3.7+ compatible
Quick Start
# 1. Install (from PyPI)
pip install jenkins-mcp-tools
# Or install from GitHub
# pip install git+https://github.com/saurprince/jenkins-mcp-server.git
# 2. Test connection
jenkins-mcp-server --test \
--url https://your-jenkins.example.com:8443 \
--username your-username \
--token your-api-token \
--no-verify-ssl
# 3. Add to Cursor (see "Cursor IDE Integration" below)
Installation
From GitHub (Recommended)
pip install git+https://github.com/saurprince/jenkins-mcp-server.git
# Or via SSH
pip install git+ssh://git@github.com/saurprince/jenkins-mcp-server.git
From Source
git clone https://github.com/saurprince/jenkins-mcp-server.git
cd jenkins-mcp-server
pip install .
Sharing with Your Team
There are several ways to distribute the server to others:
Option A: Install from Git
Anyone with access to the GitHub repo can install directly:
pip install git+https://github.com/saurprince/jenkins-mcp-server.git
Option B: Share a wheel file
Build a wheel and send the .whl file to teammates:
pip install build
python -m build --wheel
# Share dist/jenkins_mcp_server-2.0.0-py3-none-any.whl
Recipients install with:
pip install jenkins_mcp_server-2.0.0-py3-none-any.whl
Option C: Private PyPI / Artifactory
Upload the wheel to a private package index:
pip install twine
twine upload --repository internal dist/*.whl
Then anyone can pip install jenkins-mcp-server from the internal index.
Option D: Public PyPI
pip install twine
twine upload dist/*.whl
Then anyone in the world can pip install jenkins-mcp-tools.
Access Modes
The server supports three access modes, controlled via --mode flag or JENKINS_MODE environment variable:
| Mode | Tools Available | Use Case |
|---|---|---|
read-only |
20 read-only tools | Safe for analysts, viewers, dashboards |
standard (default) |
22 tools (read + trigger/stop builds) | Day-to-day developer use |
admin |
All 27+ tools (read + write + create/delete) | Jenkins administrators |
Read-only mode (safe sharing)
jenkins-mcp-server --mode read-only --url URL --username USER --token TOKEN
In read-only mode, any attempt to call trigger_build, stop_build, create_job, etc. returns a clear error message explaining which mode is required.
Standard mode (default)
jenkins-mcp-server --url URL --username USER --token TOKEN
Allows triggering and stopping builds, but not creating, deleting, or configuring jobs.
Admin mode
jenkins-mcp-server --mode admin --url URL --username USER --token TOKEN
Full access including job creation, deletion, configuration changes, folder management, and build replay.
Configuration
Environment Variables
| Variable | Description | Default |
|---|---|---|
JENKINS_URL |
Jenkins server URL | (required) |
JENKINS_USERNAME |
Jenkins username | (required) |
JENKINS_TOKEN |
Jenkins API token | (required) |
JENKINS_VERIFY_SSL |
false to disable SSL verification |
true |
JENKINS_MODE |
Access mode: read-only, standard, admin |
standard |
Getting a Jenkins API Token
- Log in to Jenkins in your browser
- Click your username (top-right) → Configure
- Scroll to API Token → Add new Token
- Name it (e.g.,
cursor-mcp-server) → Generate - Copy immediately -- it's shown only once
Cursor IDE Integration
Global Configuration
Add to ~/.cursor/mcp.json (Linux/Mac) or %USERPROFILE%\.cursor\mcp.json (Windows):
Standard mode (default for developers)
{
"mcpServers": {
"jenkins": {
"command": "jenkins-mcp-server",
"args": [
"--url", "https://your-jenkins.example.com:8443",
"--username", "your-username",
"--token", "your-api-token",
"--no-verify-ssl"
],
"disabled": false,
"timeout": 60
}
}
}
Read-only mode (safe for sharing with analysts)
{
"mcpServers": {
"jenkins": {
"command": "jenkins-mcp-server",
"args": [
"--url", "https://your-jenkins.example.com:8443",
"--username", "viewer-username",
"--token", "viewer-api-token",
"--no-verify-ssl",
"--mode", "read-only"
]
}
}
}
Admin mode (Jenkins administrators)
{
"mcpServers": {
"jenkins": {
"command": "jenkins-mcp-server",
"args": [
"--url", "https://your-jenkins.example.com:8443",
"--username", "admin-username",
"--token", "admin-api-token",
"--no-verify-ssl",
"--mode", "admin"
]
}
}
}
Using Environment Variables
{
"mcpServers": {
"jenkins": {
"command": "jenkins-mcp-server",
"args": ["--no-verify-ssl"],
"env": {
"JENKINS_URL": "https://your-jenkins.example.com:8443",
"JENKINS_USERNAME": "your-username",
"JENKINS_TOKEN": "your-api-token",
"JENKINS_MODE": "standard"
}
}
}
}
Workspace-specific Configuration
Create .cursor/mcp.json in your project root with the same format above.
After configuration, restart Cursor or reload MCP servers (Cmd/Ctrl+Shift+P → "MCP: Reload Servers").
MCP Tools Reference
Jobs & Builds (read-only)
| Tool | Description | Mode |
|---|---|---|
list_jobs |
List all jobs with status, URLs, and last build info | read-only |
get_job_details |
Get detailed job information including builds, health reports | read-only |
get_build_info |
Get build status, duration, result, and actions | read-only |
get_build_console |
Get console output with line pagination | read-only |
get_progressive_console |
Stream console output for live builds (byte offset) | read-only |
get_test_report |
Get test pass/fail counts, suites, and cases | read-only |
get_coverage_report |
Get code coverage report (requires coverage plugin) | read-only |
get_build_parameters |
Get parameters used for a build | read-only |
get_artifacts |
List build artifacts | read-only |
get_pipeline_stages |
Get Pipeline stage breakdown with status and duration | read-only |
get_job_config |
Get raw XML configuration of a job | read-only |
Infrastructure (read-only)
| Tool | Description | Mode |
|---|---|---|
get_queue_info |
Get build queue information | read-only |
get_queue_item |
Get details about a specific queue item | read-only |
list_nodes |
List all Jenkins nodes/agents with status | read-only |
get_node_info |
Get specific node details | read-only |
list_views |
List all configured views | read-only |
get_view_info |
Get view details including contained jobs | read-only |
list_credentials |
List credential IDs (values never exposed) | read-only |
get_system_info |
Get Jenkins version, executor count, mode | read-only |
list_plugins |
List installed plugins with versions | read-only |
Build Control (standard)
| Tool | Description | Mode |
|---|---|---|
trigger_build |
Trigger a new build with optional parameters | standard |
stop_build |
Stop a running build | standard |
Administration (admin)
| Tool | Description | Mode |
|---|---|---|
create_job |
Create a job from XML config or built-in template | admin |
delete_job |
Delete a job permanently | admin |
enable_job |
Enable a disabled job | admin |
disable_job |
Disable a job (prevents new builds) | admin |
copy_job |
Copy an existing job to create a new one | admin |
update_job_config |
Update a job's XML configuration | admin |
create_folder |
Create a folder for organizing jobs | admin |
replay_build |
Replay a Pipeline build with optional modified script | admin |
Job Templates (used with create_job)
When using create_job without providing raw XML, you can use a built-in template:
| Template | Description |
|---|---|
freestyle |
Basic freestyle project with shell build step |
pipeline |
Pipeline job with inline Groovy script |
pipeline-scm |
Pipeline job loading Jenkinsfile from Git SCM |
multibranch |
Multibranch Pipeline scanning branches for Jenkinsfile |
CLI Reference
# --- Read commands ---
jenkins-mcp list-jobs # List all jobs
jenkins-mcp list-jobs -f simple # Simple format
jenkins-mcp list-jobs --folder MyFolder # Jobs in a folder
jenkins-mcp get-job MyJob # Job details
jenkins-mcp get-build --job MyJob --build 64 # Build info
jenkins-mcp get-console --job MyJob --build 64 --tail 50 # Console output
jenkins-mcp get-test-report --job MyJob --build 64 -s # Test summary
jenkins-mcp get-coverage --job MyJob --build 64 # Coverage report
jenkins-mcp get-stages --job MyJob --build 64 # Pipeline stages
jenkins-mcp get-config MyJob # Job XML config
jenkins-mcp system-info # Jenkins system info
jenkins-mcp list-plugins # Installed plugins
jenkins-mcp list-views # All views
jenkins-mcp list-credentials # Credential IDs
jenkins-mcp list-nodes # Nodes/agents
jenkins-mcp get-queue # Build queue
# --- Build control ---
jenkins-mcp trigger-build MyJob # Trigger build
jenkins-mcp trigger-build MyJob -p '{"BRANCH":"main"}' # With parameters
jenkins-mcp stop-build --job MyJob --build 123 # Stop build
# --- Admin commands ---
jenkins-mcp create-job NewJob --template pipeline --script 'pipeline { agent any { stages { stage("Build") { steps { echo "hi" } } } } }'
jenkins-mcp create-job NewJob --config-file config.xml # From XML file
jenkins-mcp create-job NewJob --template pipeline-scm --repo-url https://github.com/user/repo.git
jenkins-mcp delete-job OldJob -y # Delete (skip confirm)
jenkins-mcp enable-job MyJob # Enable
jenkins-mcp disable-job MyJob # Disable
jenkins-mcp copy-job SourceJob NewCopy # Copy job
jenkins-mcp update-config MyJob new-config.xml # Update config
jenkins-mcp create-folder MyFolder # Create folder
jenkins-mcp replay-build --job MyJob --build 64 # Replay pipeline
Example Prompts for Cursor
Once configured, you can ask Cursor things like:
Monitoring:
- "List all Jenkins jobs and their status"
- "Show me the pipeline stages for MyJob build 64"
- "What plugins are installed on Jenkins?"
- "Get the system info for our Jenkins instance"
Debugging:
- "Get the test report for MyJob build 64 -- what failed?"
- "Show me the last 100 lines of console output for MyJob build 64"
- "What parameters were used for build 64?"
- "Get the coverage report for the latest build"
Build control:
- "Trigger a new build for MyJob with BRANCH=feature-x"
- "Stop build 123 of MyJob"
Administration (admin mode):
- "Create a new pipeline job called 'deploy-staging' with this Groovy script: ..."
- "Copy the 'nightly-tests' job to 'nightly-tests-v2'"
- "Disable the 'old-job' job"
- "Create a folder called 'team-alpha' for organizing jobs"
Python API
from jenkins_mcp_server import JenkinsClient
client = JenkinsClient(
url="https://your-jenkins.example.com:8443",
username="your-username",
token="your-api-token",
verify_ssl=False,
)
# List jobs
jobs = client.list_jobs()
for job in jobs:
print(f"{job['name']}: {job.get('color', 'unknown')}")
# Build info
build = client.get_build_info("MyJob", 64)
print(f"Result: {build.get('result')}, Duration: {build.get('duration')}ms")
# Pipeline stages
stages = client.get_pipeline_stages("MyPipeline", 10)
for stage in stages.get('stages', []):
print(f" {stage['name']}: {stage['status']}")
# System info
info = client.get_system_info()
print(f"Jenkins {info['version']}, {info['numExecutors']} executors")
# Trigger a build
result = client.trigger_build("MyJob", {"BRANCH": "main"})
print(f"Queued: {result['queue_url']}")
# Create a job from template
from jenkins_mcp_server.client import render_job_template
config = render_job_template("pipeline", script="pipeline { agent any { stages { stage('Build') { steps { echo 'hello' } } } } }")
client.create_job("my-new-pipeline", config)
Troubleshooting
MCP Server Not Loading in Cursor
- Verify installation:
pip show jenkins-mcp-tools - Test connection:
jenkins-mcp-server --test --url URL --username USER --token TOKEN --no-verify-ssl - Check Cursor's Output panel for MCP errors
- Restart Cursor after changing
mcp.json
SSL Certificate Errors
Use --no-verify-ssl flag or set JENKINS_VERIFY_SSL=false.
Authentication Errors (401)
- Verify username and token are correct
- Ensure the API token hasn't been revoked
- Check if your account has API access permissions
- Regenerate the token from Jenkins (User → Configure → API Token)
Mode-related Errors
If you see "Tool 'X' requires 'admin' mode but server is running in 'standard' mode":
- Restart the server with
--mode adminto enable admin tools - Or set
JENKINS_MODE=adminin your environment / Cursor config
Connection Timeouts
- Ensure VPN is connected (if required)
- Test:
curl -I https://your-jenkins.example.com:8443
Changelog
v2.0.0
- Access modes:
read-only,standard,adminwith--modeflag andJENKINS_MODEenv var - 15 new MCP tools: pipeline stages, job config, credentials, views, system info, plugins, job CRUD, folder management, build replay
- Job templates: Create freestyle, pipeline, pipeline-scm, and multibranch jobs from built-in templates
- CLI expansion: New commands for all new tools
- Sharing guide: Documentation for distributing to teammates
v1.0.0
- Initial release with 12 MCP tools
- CLI tool with subcommands
- Cursor IDE integration
Contributing
- Clone the repository
- Install dev dependencies:
pip install -e ".[dev]" - Run tests:
pytest tests/ - Create a feature branch and merge request
License
MIT License - See LICENSE file for details.
Support
For issues or questions, open an issue on the repository.
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。
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。
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。
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。