Hey everyone,

I recently participated in Quira Quest 19. For Challenge 1, I built a basic CLI app that connects with MindsDB's Knowledge Bases and Agents.

Here's the repo:
https://github.com/pheonix-coder/kb-manager

It's a small CLI app using Typer. Under the hood, it just runs SQL queries against MindsDB. I kept it modular, so people can reuse the commands and plug them into other apps or scripts. Just enough to show how the pieces fit together in a real system.

You can do the following from the CLI:

• Create a knowledge base (KB) with metadata and embedding configs
• Ingest CSV data
• Run semantic search queries
• Summarize the KB using OpenAI (GPT-3.5)
• Create a chat agent connected to the KB
• Chat with the agent and get responses

All the setup steps are in the README:
https://github.com/pheonix-coder/kb-manager#️-setup

---

Why I Built It

Mostly to get a feel for how MindsDB actually handles knowledge and agents behind the scenes. The docs give you the high-level picture, but I wanted to run stuff directly and see the full flow: from raw data to intelligent agent responses.

I didn't want to build a fancy frontend. Just CLI + some color output and animations. Something you can fire up in a terminal and see results fast.

---

What This Is Useful For

• You want to test the KB → Embedding → Search pipeline
• You’re debugging prompts or trying different embedding models
• You want to show someone how this stuff works (dev, investor, teammate)
• You want to build something minimal to summarize text datasets
• You just want to learn MindsDB quickly and don’t like clicking through dashboards

---

Realistic Use Cases

1. Internal KB from CSV/Docs
Let’s say you have a bunch of customer quotes or logs. You can just ingest them and start querying with natural language. No dashboard, no API keys to manage — just the CLI.

2. Search Your Data in Plain English
Run query-kb and type something like:
ways to live life
You’ll get results based on semantic similarity. (Like Einstein quotes.)

3. Summarize Big Datasets
Run summarize-kb on a 2,000-row CSV, and get a 2-line summary. Useful if you don’t want to read through the whole thing.

4. Agents That Actually Know Your Data
Create a chatbot that only responds based on the KB content.
So you can ask: “What did Einstein say about imagination?” — and it only pulls from your dataset.

5. Demo or Teaching Tool
This CLI is good for showing how vector search and agents work without spinning up a web app. Great for product teams, new devs, or anyone learning LLMs + vector DBs.

---

Demo Video

If you want to see it in action, here’s a quick demo:
https://www.youtube.com/watch?v=K-cufogVz0Q

---

That’s it. Not a big project, but helped me understand a lot.

If this kind of thing interests you, follow me here:
https://x.com/pheonix_coder