Supervisor agent pattern in Agentic AI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Role of the Supervisor Agent in the Supervisor Agent Pattern

In the supervisor agent pattern, what is the primary role of the supervisor agent?

A To directly execute all tasks without delegation

B To monitor and coordinate subordinate agents to ensure task completion

C To replace subordinate agents when they fail

D To generate raw data for training machine learning models

💡 Hint

Think about who manages and oversees other agents in a team.

The supervisor agent oversees and coordinates subordinate agents, ensuring tasks are completed correctly and efficiently.

❓ Model Choice

intermediate

Choosing the Best Model for a Supervisor Agent

Which type of model is most suitable for implementing a supervisor agent that needs to evaluate multiple subordinate agents' outputs and decide the next steps?

A A rule-based decision system with conditional logic

B A simple linear regression model

C A convolutional neural network (CNN) designed for image recognition

D A recurrent neural network (RNN) for sequence prediction

💡 Hint

Consider the need for clear decision rules rather than pattern recognition.

Supervisor agents often use rule-based systems to evaluate outputs and decide actions based on conditions, rather than complex pattern recognition models.

❓ Predict Output

advanced

Output of Supervisor Agent Coordination Code

What is the output of the following Python code simulating a supervisor agent coordinating two subordinate agents?

Agentic AI

class SubordinateAgent:
    def __init__(self, name):
        self.name = name
    def perform_task(self):
        return f"Task done by {self.name}"

class SupervisorAgent:
    def __init__(self, agents):
        self.agents = agents
    def coordinate(self):
        results = []
        for agent in self.agents:
            results.append(agent.perform_task())
        return results

agents = [SubordinateAgent('Agent1'), SubordinateAgent('Agent2')]
supervisor = SupervisorAgent(agents)
output = supervisor.coordinate()
print(output)

A ['Task done by Agent1']

B ['Task done by Agent2', 'Task done by Agent1']

C TypeError: 'SubordinateAgent' object is not iterable

D ['Task done by Agent1', 'Task done by Agent2']

💡 Hint

Look at the order of agents in the list and how results are collected.

The supervisor calls perform_task on each agent in order, collecting their results in a list.

❓ Hyperparameter

advanced

Key Hyperparameter for Supervisor Agent's Decision Threshold

In a supervisor agent that decides to reassign tasks based on confidence scores from subordinate agents, which hyperparameter is critical to tune?

A Confidence threshold for task reassignment

B Number of layers in the supervisor agent's neural network

C Batch size used during training of subordinate agents

D Learning rate of the subordinate agents' models

💡 Hint

Focus on the parameter that controls when the supervisor decides to act.

The confidence threshold determines when the supervisor decides a subordinate agent's output is unreliable and needs reassignment.

🔧 Debug

expert

Debugging Supervisor Agent Coordination Failure

Given the following code snippet for a supervisor agent, what error will occur when running it and why? class SupervisorAgent: def __init__(self, agents): self.agents = agents def coordinate(self): results = [] for agent in self.agents: results.append(agent.perform_task) return results class SubordinateAgent: def perform_task(self): return "done" agents = [SubordinateAgent(), SubordinateAgent()] supervisor = SupervisorAgent(agents) output = supervisor.coordinate() print(output)

A TypeError: 'method' object is not callable

B ['done', 'done']

C ['<bound method SubordinateAgent.perform_task of <__main__.SubordinateAgent object at 0x...>>', '<bound method SubordinateAgent.perform_task of <__main__.SubordinateAgent object at 0x...>>']

D AttributeError: 'SubordinateAgent' object has no attribute 'perform_task'

💡 Hint

Check if the method is called or just referenced.

The code appends the method itself without calling it, so the output is a list of method references, not their return values.