QBot: A CLI MCP Client to support DevOps

I wanted to give it a try with OpenAI Codex or Claude Code. I found this 1-dollar promo code for OpenAI, hence, I went with Codex.

I have to say, I am quite impressed. I asked it to generate a Claude code and Codex clone, and it did a pretty good job for a first attempt and MVP. I am aiming to create QBot, a CLI MCP client that can interact with any MCP server, but mainly focused on supporting DevOps practitioners to interact with their infrastructure all via API calls; thanks to HAPI MCP Stack, this is now possible.

Let me show you the MVP generated this Sunday afternoon.

I can connect to an MCP server. List all available tools. Call a tool directly, but I prefer to leverage natural language through a prompt. I can configure the LLM provider and the model to use.

With that, I can now interact with my MCP server and get things done. I tested with the Petstore example. If you have been following my blog and videos, you know I am using it a lot for demos, because it is simple and available everywhere.

The Demo (MVP in Action)

In the demo video, I walk through the very first working version of QBot:

1. Listing tools: Seeing what actions the MCP server provides.

2. Checking configs: Confirming Ollama 3 is running and connected.

3. Executing requests: Running get pet by ID 1 or deleting an order with ID 10 — all powered by natural language mapped to real API calls.

It may look simple now, but that’s exactly the point: accessibility. A beginner DevOps engineer could get the same results as a seasoned pro, without memorizing commands.

Start QBot, connect to the MCP server, and list all available tools!

$ qbot --url http://localhost:3000/mcp
QBot • MCP REPL  Type text to chat, /help for commands
➜ Connected to http://localhost:3000/mcp
qbot> /tools
... removed for brevity ...  
qbot> /llm status
Provider: ollama
Model: llama3.1:latest
Keys: (none)
Base URLs: ollama
qbot> get pet with ID 1
➜ Calling tool getPetById
assistant: The pet with ID 1 has the following details:

* Category: Cats (ID 2)
* Name: Cat 1
* Photo URLs: url1, url2
* Tags: tag1, tag2
* Status: available
qbot> delete order with ID 10
➜ Calling tool deleteOrder
assistant: Order with ID 10 has been deleted successfully.
qbot> /exit

You can see how easy it is to interact with the MCP server and get things done. I can now leverage the power of LLMs to interact with my infrastructure and get things done.

I am planning to add more features, like:

• Support for more advanced prompts, like few-shot learning, etc.

• Support for more advanced features, like tool chaining, etc.

Stay tuned, and let me know what you think. I am open to feedback and suggestions.

Be HAPI. 😁 Go Rebels! ✊🏽