Skip to main content
In this lesson, we’ll show you how to run Flask tests and configure different AI models inside Cursor for autocompletion, code explanation, or multi-step tasks. By the end, you’ll know how to select, customize, and manage models to fit your development workflow.

​
1. Running Flask Application Tests

Start by defining fixtures and writing basic tests for your Flask app: 
import os
import pytest
from app import app, init_db

@pytest.fixture
def client():
    # Initialize test client and database
    with app.test_client() as client:
        with app.app_context():
            init_db()
        yield client
    # Clean up temp database files
    os.close(db_fd)
    os.unlink(app.config['DATABASE'])

def test_index_redirect(client):
    """Ensure the index page redirects to login."""
    response = client.get('/')
    assert response.status_code == 302
    assert '/login' in response.headers['Location']

def test_register(client):
    """Verify user registration flow."""
    response = client.post(
        '/register',
        data={'username': 'testuser', 'password': 'testpass'},
        follow_redirects=True
    )
    assert response.status_code == 200
    assert b'Account created successfully!' in response.data

​
2. Browsing Model Options in Cursor

Open the chat interface to explore AI model settings:
  • Left sidebar: Toggle autocompletion modes (Ask, Agentic, Edit).
  • Right pane: Choose a specific model or leave on Auto-select.
The image shows a code editor with a dropdown menu for selecting an AI model, highlighting "claude-3.5-sonnet." The background displays Python code related to a web application.
Auto-select lets Cursor pick the best model based on your preferences and credits.

​
3. AI Model Categories

CategoryDescription
Traditional LLMDirect response (e.g., ChatGPT, Claude).
Thinking LLMReturns answer + step-by-step reasoning.
Agentic LLMAutomates complex, multi-step workflows (Agentic mode).

​
Example: 2 + 2

Traditional LLM 
Prompt: "What is 2 plus 2?"
Response: "4"
Thinking LLM 
To find 2 plus 2, I add the numbers together: 2 + 2 = 4. Therefore, the answer is 4.
ModelKey StrengthPricing Example
Claude 3.5 SonnetReliable baseline—
Claude 3.7Advanced reasoning$0.05 per request/tool
GPT-4High speed & accuracy—
GPT-4 MiniCost-effective—
Costs can add up quickly—monitor your credits and usage in Cursor Pro or via public API billing.

​
5. Ask Mode for Detailed Explanations

Switch to Ask Mode to request human-friendly explanations:
“Explain what def test_index_redirect means.” 
def test_index_redirect(client):
    """Test that the index page redirects to login."""
    response = client.get('/')
    assert response.status_code == 302
    assert '/login' in response.headers['Location']
With GPT-4 in Ask Mode, you might see:
“This function uses Flask’s test client to ensure a request to / returns a 302 redirect to /login.”
Experiment by selecting a thinking model and adding prompts like “explain it like I’m five” for varied results.

​
6. Customizing Models in Settings

Under Cursor Settings → Models, you can:
The image shows a code editor interface with a "Cursor Settings" panel open, displaying options for entering API keys for OpenAI, Anthropic, Google, and Azure. The left sidebar lists project files, and a terminal is visible at the bottom.
Enable or disable the models you need, then close the panel to apply changes.
The image shows a code editor with a file directory on the left and a "Cursor Settings" window in the center, listing various model names. A terminal is open at the bottom, and a chat-like explanation is visible on the right.

Watch Video