home / mcp / argocd mcp server
Provides ArgoCD capabilities via MCP for AI-assisted, multi-cluster GitOps control and deployment management.
Configuration
View docs{
"mcpServers": {
"asklokesh-argocd-mcp-server": {
"command": "python",
"args": [
"-m",
"argocd_mcp.server"
],
"env": {
"ARGOCD_SERVER": "argocd.example.com",
"ARGOCD_TIMEOUT": "30",
"ARGOCD_GRPC_WEB": "true",
"ARGOCD_INSECURE": "false",
"ARGOCD_AUTH_TOKEN": "your_auth_token",
"ARGOCD_CLIENT_KEY": "/path/to/key.pem",
"ARGOCD_CLIENT_CERT": "/path/to/cert.pem"
}
}
}
}You can run an MCP server that connects ArgoCD with GenAI-powered workflows to manage Kubernetes deployments. This MCP server exposes ArgoCD capabilities through the MCP protocol, enabling intelligent, automated GitOps operations across multiple clusters and environments.
You will run the MCP server locally or in your platform and connect it with an MCP client. The server acts as a bridge to ArgoCD, letting you list, create, synchronize, and monitor applications, manage projects and repositories, register clusters, and drive CI/CD workflows with AI-assisted automation. You can configure multi-cluster environments, progressive delivery strategies, disaster recovery, and RBAC-enabled access control. Use the provided tools to perform common GitOps tasks and let the AI-enhanced flows handle complex deployment orchestration.
To start, install the server, configure your ArgoCD connection, and run the Python module that serves the MCP endpoints. Then, use your MCP client to invoke the available tools for specific operations such as application management, project setup, and health checks.
# Prerequisites: Python 3.9+ and pip must be available on your system.
# Install the MCP server package from PyPI
pip install argocd-mcp-server
# Or install from source if you prefer local development
# git clone https://github.com/LokiMCPUniverse/argocd-mcp-server.git
# cd argocd-mcp-server
# pip install -e .
# Run the MCP server (example using the CLI runner via Python module shown in client config)
python -m argocd_mcp.serverConfiguration uses environment variables. The following variables are used to connect to ArgoCD and control optional features. You can place these in a .env file or export them in your environment.
# ArgoCD Connection
ARGOCD_SERVER=argocd.example.com
ARGOCD_AUTH_TOKEN=your_auth_token
ARGOCD_INSECURE=false
# Optional Settings
ARGOCD_GRPC_WEB=true
ARGOCD_TIMEOUT=30
ARGOCD_CLIENT_CERT=/path/to/cert.pem
ARGOCD_CLIENT_KEY=/path/to/key.pem
# Multi-Cluster Support (example for multiple clusters)
ARGOCD_PROD_SERVER=argocd-prod.example.com
ARGOCD_PROD_TOKEN=prod_token
ARGOCD_DEV_SERVER=argocd-dev.example.com
ARGOCD_DEV_TOKEN=dev_tokenThe server supports running against multiple ArgoCD instances, progressive delivery strategies, disaster recovery workflows, auditing, and secret management integration. You can configure multiple clusters with distinct auth tokens and default namespaces, enable canary or blue-green deployment patterns, and automate rollbacks on failures.
If you encounter issues, verify your ArgoCD connection details, ensure the tokens have appropriate permissions, and check that the MCP server process is running and reachable by your MCP client. Review logs for authentication errors, network connectivity, and RBAC misconfigurations.
Follow standard security practices for GitOps integrations: use service accounts with minimal permissions, enable RBAC, apply network policies, and enable audit logging. Rotate tokens regularly and protect secrets using sealed secrets or a secret operator.
The server returns structured error information for failed operations. When an operation fails, inspect the error code and message to determine the root cause, such as authentication problems, permission issues, or misconfigurations in the ArgoCD setup.
Optimize performance by using application sets for scalable deployments, enabling caching where supported, and configuring synchronization windows to control deployment timing. Prefer webhook-driven updates over polling to reduce load.
You can combine GitOps workflows with AI-assisted decision making to automate repetitive deployments, implement canary or blue-green delivery patterns, and orchestrate multi-cluster deployments from a single control plane.
Contributions are welcome. The project is released under the MIT License. Follow typical contribution guidelines and submit pull requests with clear motivation and test coverage.
The MCP server exposes a set of end-user tools to manage applications, projects, repositories, clusters, and more. You can list, create, sync, rollback, and query status for applications; create and configure projects; add repositories and Helm charts; manage clusters; and work with ApplicationSets and health data.
List applications with optional project and namespace filters.
Create a new ArgoCD application with repository, source, destination, and sync policy.
Sync an application with options for pruning, forcing, and strategy hooks.
Retrieve the current status of an application, with optional refresh.
Rollback an application to a previous revision.
Create a new ArgoCD project with source repos, destinations, and resource whitelists.
Add a Git repository or Helm repository to ArgoCD.
Add a Helm chart repository to ArgoCD configuration.
Register a new Kubernetes cluster with connection details.
Create an ApplicationSet for template-based multi-cluster deployments.
Get health status for a specific application.
Fetch the resource tree for an application to inspect its manifests.