Argo Workflow

You can deploy this MCP server to let AI agents manage Argo Workflows through the Argo Server REST API. It supports creating and listing templates, submitting and tracking workflow instances, and secure access with tokens and optional TLS verification, all via MCP-compatible endpoints.

How to use

Connect your MCP client to the Argo Workflow MCP Server using either a remote Argo Server URL or a local setup through port-forwarding. You can manage Workflow Templates and Workflow instances, monitor statuses, and perform cleanups with standard MCP actions.

Two common connection patterns are available. Use the HTTP method to connect to a remote Argo Server when you have a reachable URL and a valid token. Use the STDIO method to run the MCP server locally (via npx) and point it at your Argo Server, optionally skipping TLS verification for development or internal environments.

Typical workflows you can perform include: create, query, list, and delete Workflow Templates; submit, monitor, list, and delete Workflows. Your MCP client can call these endpoints to integrate Argo operations into automated agents or higher-level orchestration flows.

How to install

Prerequisites: you need Node.js installed on your system. You can install Node.js from the official site or use your system package manager.

Option 1: Use npx (no installation required) to run the MCP server directly against your Argo Server. This is ideal for quick starts and ad-hoc usage.

Option 2: Install globally so you can run the MCP server from anywhere.

Complete steps below to set up both options.

Other important notes

Configuration requires you to specify the Argo Server address and, if you are using a secured Argo Server, the access token. You can also disable TLS certificate verification with an explicit flag for development or internal environments. For production, always use valid TLS certificates.

AI integration example shows how to configure a client (Kiro) to run the MCP server as a domain-specific action provider for Argo workflows.

Troubleshooting covers common TLS certificate errors, connectivity issues, and permission checks to ensure the MCP client can call Argo APIs successfully.

Kiro configuration example

{
  "mcpServers": {
    "argo-workflow": {
      "command": "npx",
      "args": [
        "-y",
        "argo-workflow-mcp-server",
        "--argo-server",
        "https://argo-server.example.com",
        "--argo-token",
        "your-auth-token",
        "--namespace",
        "argo"
      ]
    }
  }
}

Security and access

Protect access with tokens and TLS. Use --argo-token for Argo authentication and --argo-insecure only in development or trusted internal environments to skip TLS verification. Do not use insecure mode in production.

Examples and troubleshooting

If you encounter a self-signed certificate error, enable insecure TLS verification during development. If you cannot connect, verify the Argo Server URL, confirm token validity, and check network connectivity. Ensure your ServiceAccount has sufficient RBAC permissions for the MCP operations.

Available tools

create_workflow_template

Create a new Workflow Template in Argo, enabling you to store reusable workflow blueprints.

get_workflow_template

Fetch details of a specific Workflow Template by name.

list_workflow_templates

List all available Workflow Templates in the configured Argo namespace.

delete_workflow_template

Delete a specified Workflow Template from Argo.

submit_workflow

Submit a new Workflow instance from a Workflow Template and start its execution.

get_workflow_status

Query the current status and progress of a running Workflow.

list_workflows

List all Workflows in the target namespace.

delete_workflow

Delete a specific Workflow from Argo.