MCP Argo Server

What is MCP Argo Server?

MCP Argo Server is a lightweight CLI tool that wraps Argo Workflows, providing an MCP-compliant JSON-RPC interface over STDIN/STDOUT for launching workflows, checking status, and retrieving results. It is written in Go and uses Foxy Contexts and client-go to interact with Kubernetes and Argo resources.

How to use MCP Argo Server?

Open the project in a dev container or clone and run go mod tidy. Use make cluster to install a k3d cluster and set up Argo. Then run make run to start the MCP server. Alternatively, use the included Python test client by running make install and python test_with_autogen.py in the python/ directory.

Key features of MCP Argo Server

• MCP-compliant server for Argo Workflows
• JSON-RPC communication over STDIN/STDOUT
• Launch workflows, check status, retrieve results
• Lightweight CLI tool written in Golang
• Integrates with Kubernetes via client-go
• Includes Python test client for demonstration

Use cases of MCP Argo Server

• Automating Argo Workflow submission from MCP clients
• Monitoring workflow status programmatically
• Integrating workflow management into AI agents or tools
• Testing and debugging Argo workflows in a dev container

FAQ from MCP Argo Server

What is the runtime requirement?

It runs inside a development container with pre-installed dependencies, or locally with Go. It requires a k3d cluster and Argo installation.

How does the server communicate?

It uses JSON-RPC over STDIN/STDOUT, leveraging Foxy Contexts for RPC handling.

Where are workflow data stored?

Workflows are managed by Argo Workflows on a Kubernetes cluster.

What transport does the server use?

The server uses STDIN/STDOUT transport for MCP.

Is there an authentication mechanism?

The README does not mention authentication; it is intended for local development or within a trusted environment.