Task CRUD API (FastAPI)

A lightweight RESTful CRUD API built with Python and FastAPI for managing to-do tasks.

Built for Week 2 of the AI Fluency backend track to demonstrate:

• HTTP request-response cycles
• HTTP status codes
• Input validation
• RESTful CRUD operations
• Automated OpenAPI documentation with Swagger UI

---

How to Run Locally

Prerequisites

• Python 3.10+
• pip package manager

1. Install Dependencies

pip install fastapi uvicorn pydantic

2. Start the Server

python -m uvicorn main:app --reload

Base URL: http://127.0.0.1:8000

Interactive Swagger UI: http://127.0.0.1:8000/docs

---

API Endpoints

Method	Endpoint	Description	Expected Body	Success Status	Error Status
GET	/	API Metadata & Routes	None	200 OK	-
GET	/health	Server Health Status	None	200 OK	-
GET	/stats	Task Counts (Total/Done/Open)	None	200 OK	-
GET	/tasks	List Tasks (Optional done/search)	None	200 OK	-
GET	/tasks/{id}	Get Single Task by ID	None	200 OK	404 Not Found
POST	/tasks	Create New Task	{"title": "Buy milk"}	201 Created	400 Bad Request
PUT	/tasks/{id}	Update Task	{"title": "New", "done": true}	200 OK	400 Bad Request, 404 Not Found
DELETE	/tasks/{id}	Delete Task	None	204 No Content	404 Not Found
POST	/reset	Reset DB to 3 initial seed items	None	200 OK	-

---

Terminal Verification

The API can be tested from the terminal using curl.

Create a Task

curl -i -X POST http://127.0.0.1:8000/tasks \
  -H "Content-Type: application/json" \
  -d '{"title":"Complete Week 2 Assignment"}'

Expected response:

HTTP/1.1 201 Created
date: Sun, 16 Aug 2026 17:00:00 GMT
server: uvicorn
content-type: application/json

{
  "id": 4,
  "title": "Complete Week 2 Assignment",
  "done": false
}

---

In-Memory Persistence Experiment

Observation

When new tasks are created and the Uvicorn server is restarted using:

Ctrl+C

followed by:

python -m uvicorn main:app --reload

all newly created tasks disappear, and the application returns to the initial 3 seed tasks.

Why Does This Happen?

The application stores its data in a volatile Python list called:

tasks_db

This list exists only in the computer's RAM (memory) while the Python process is running.

When the server process stops:

1. The Python process terminates.
2. The tasks_db list is destroyed.
3. All newly created tasks are lost.
4. When the server starts again, the original seed data is created.

This demonstrates why real-world applications require persistent storage such as:

• SQL databases
• NoSQL databases
• Other persistent storage engines

---

Stage 7 — AI Rematch ("AI vs Me")

Full Prompt Used

Write a single-file Python REST API using FastAPI that manages an in-memory task to-do list. Implement GET /, GET /health, GET /tasks, GET /tasks/{id} (404 if missing), POST /tasks (generates next ID, validates non-empty title returning 400 Bad Request, sets default done=False), PUT /tasks/{id} (updates title/done, returns 400 or 404), and DELETE /tasks/{id} (returns 204 No Content). Ensure Swagger UI is available at /docs.

---

Code Comparison & Findings

What the AI Did Better

The AI used Pydantic's:

Field(min_length=1)

for declarative validation directly inside the request model.

This provides a clean way to prevent completely empty strings from being submitted.

---

What the AI Got Wrong / Ignored

1. Whitespace Bypass

The AI accepted whitespace-only strings such as:

"   "

as valid task titles.

Although min_length=1 rejects an actually empty string, whitespace characters still count toward the length.

Fix

Manual string stripping was required:

payload.title.strip()

This ensures that a title containing only whitespace is rejected.

---

2. HTTP 204 Protocol Violation

The AI returned:

{
  "message": "Task deleted"
}

alongside:

204 No Content

This is incorrect.

A 204 No Content response must not contain a response body.

Correct Behavior

The DELETE endpoint should return only:

204 No Content

with no JSON response body.

---

What the Prompt Was Missing

The original prompt did not explicitly specify how new task IDs should be generated.

The AI used:

len(tasks_db) + 1

This can create duplicate IDs.

Example

Suppose the database contains:

1
2
3

If task 3 is deleted, the database now contains:

1
2

Using:

len(tasks_db) + 1

would generate:

3

which happens to work in this case.

But if an earlier ID is deleted:

1
3
4

then:

len(tasks_db) + 1

would produce:

4

which is already being used.

Fix

The ID-generation logic was updated to:

max(existing_ids) + 1

This generates the next ID based on the highest existing ID rather than the number of tasks currently stored.

---

Key Learning Outcomes

This project demonstrates practical backend concepts including:

• REST API architecture
• FastAPI routing
• HTTP methods
• HTTP status codes
• CRUD operations
• Pydantic validation
• Swagger / OpenAPI documentation
• In-memory data storage
• API testing with curl
• Server lifecycle and data persistence
• AI-generated code evaluation
• Identifying and fixing AI coding mistakes

---

Project Status

Week 2 — AI Fluency Backend Track

Technology: Python + FastAPI + Pydantic + Uvicorn

Storage: In-memory Python list

Documentation: OpenAPI / Swagger UI