Cleric Query Agent Assignment

Introduction

This document outlines the requirements and guidelines for the Cleric Query Agent Assignment. Your task is to develop an AI agent capable of accurately answering queries about applications deployed on a Kubernetes cluster.

Objective

Create an AI agent that interacts with a Kubernetes cluster to answer queries about its deployed applications.

Assignment Details

Technical Requirements

• Use Python 3.10
• The kubeconfig file will be located at ~/.kube/config
• Utilize GPT-4 or a model with comparable performance for natural language processing

API Specifications

Your agent should provide a POST endpoint for query submission:

• URL: http://localhost:8000/query
• Port: 8000
• Payload format:

  {
      "query": "How many pods are in the default namespace?"
  }

• Response format (using Pydantic):

  from pydantic import BaseModel
  
  class QueryResponse(BaseModel):
      query: str
      answer: str

Scope of Queries

• Queries will require only read actions from your agent
• Topics may include status, information, or logs of resources deployed on Minikube
• Answers will not change dynamically
• Approximately 10 queries will be asked
• Queries are independent of each other
• Return only the answer, without identifiers (e.g., "mongodb" instead of "mongodb-56c598c8fc")

Submission Guidelines

Submit your repository to submission link

• The validator will return your score within a few minutes
• Use logging if you want to check your outputs, make sure write logs to agent.log
• If you encounter errors, wait a few minutes before retrying
• Do not refresh the browser to avoid losing your session
• Make sure to note your Submission ID for the Google form for the final submission.

Submission Requirements

1. GitHub Repository
   • Include a README.md file describing your approach
   • Ensure your main script is named main.py
2. Loom Video
   • Keep it informal and personal
   • Focus on your motivation and background
3. Submit the Loom video and submission ID for the final submission on this Google Form Link

Submission Deadline:

There is no specific deadline for submitting this assignment; however, we expect it to be completed within a reasonable amount of time.

• We understand that personal and professional responsibilities can take priority, and we encourage you to balance this assignment with your other commitments.
• Please aim to submit your work once you feel confident in your solution and it aligns with the objectives.

Testing Your Agent

We recommend testing your agent locally before submission:

1. Install Minikube
2. Set up a local Kubernetes cluster
3. Deploy sample applications
4. Run your agent and test with sample queries

Evaluation Criteria

• Accuracy of answers
• Code quality and organization
• Clarity of explanation in README and video

Example Queries and Responses

1. Q: "Which pod is spawned by my-deployment?" A: "my-pod"
2. Q: "What is the status of the pod named 'example-pod'?" A: "Running"
3. Q: "How many nodes are there in the cluster?" A: "2"