VibeLeet is a competitive prompt engineering platform where you write natural language prompts to solve coding challenges. AI generates code from your prompts, and automated tests score your solutions.

How does prompt engineering scoring work?

Your score is based on three factors: correctness (how many test cases pass), code efficiency, and prompt quality. The goal is to write the most concise and effective prompt possible.

VibeLeet 1.0 AvailableStart solving

Prompting is the next
decade of engineering.

Coding is changing. The future belongs to those who direct the machine, not just type the syntax. VibeLeet is the premier training platform to hone your architectural intuition and prompt engineering skills.

Start Solving View Leaderboard

Trusted by 4 engineers

Execution at the edge.

Write plain English. AI writes Python. Secure cloud sandboxes run the tests. Prove your ability to architect through natural language.

Write the Vision

Describe your logic, edge cases, and architectures natively in English. Treat the editor like a lead engineering whiteboard.

Instant Generation

Your prompt instantly compiles into functional Python within milliseconds, powered by industry-leading Llama 3 models.

Secure Sandboxes

Generated code is deployed into E2B's isolated execution environments running rigorous public and hidden unit tests.

solution.py

1def two_sum(nums: List[int], target: int) -> List[int]:

2 """

3 AI Generated: O(n) time, O(n) space

4 """

5 seen = {}

6 for i, num in enumerate(nums):

7 diff = target - num

8 if diff in seen:

9 return [seen[diff], i]

10 seen[num] = i

11 return []

E2B Sandbox

$ python -m pytest test_solution.py

============= test session starts =============

test_solution.py .... [100%]

============= 4 passed in 0.02s =============

How it works.

The loop is simple, but mastery takes practice. Our platform trains you to think like a staff engineer.

Read the Spec

Analyze the problem requirements, expected inputs, and edge cases to truly understand the architectural goal before writing a word.

Write the Prompt

Draft concise, unambiguous English instructions inside our Monaco editor. Structure your logic clearly for the LLM to interpret.

Execute & Score

We stream the code, test it instantly in an isolated sandbox, and award points based on test coverage and execution efficiency.

The Paradigm Shift

"Syntax is cheap.
Architecture is hard."

Historically, engineers proved their worth by memorizing API signatures and syntax quirks. Today, the machine writes the code. Your value is determined by your ability to think systematically and communicate logical models seamlessly.

VibeLeet isn't about teaching you to prompt a chatbot. It's about training you to become an elite technical director for autonomous software agents.

Real-World Scenarios

Test your logic against everything from fixing obscure RegEx bugs to architecting scalable system modules.

Bug HuntEasy

Fix the Racing Condition

This async function is dropping data. Write a prompt to identify the race condition and return a thread-safe implementation without using locks.

150 pts

Solve Now

System DesignHard

Rate Limiter Middleware

Design a Redis-backed sliding window rate limiter. Instruct the AI on exactly how to handle edge cases, precise millisecond concurrency, and fallback logic.

500 pts