Coding Platform Study Solver

This project is a local Chrome extension plus Python backend that helps you study supported coding-platform problems from inside the browser.

What it does:

• detects when a supported problem page is open
• extracts the visible problem statement and starter code from the current tab
• sends that data to a local backend
• generates a draft solution
• shows the result inside the extension popup
• lets you copy the generated code and notes manually

Important limitation:

• This project is a study assistant. It does not auto-submit solutions or claim hidden tests passed.
• It does not inject an in-page control panel anymore. The coding site page is left alone and the extension popup is the UI.
• It does not include stealth, anti-detection bypass, or any attempt to evade platform safeguards.

Supported Platforms

• LeetCode
• HackerRank
• CodeChef
• Codeforces
• AtCoder

The extension uses a platform-adapter registry in extension/platforms.js, so more sites can be added cleanly.

Project Layout

• extension/: Chrome extension files
• backend/: local Python backend and tests
• run_backend.cmd: starts the backend using a known local Python interpreter
• run_backend_tests.cmd: runs backend verification tests

Quick Start

1. Start the backend:

run_backend.cmd

2. Load the Chrome extension:

• Open chrome://extensions
• Enable Developer Mode
• Click Load unpacked
• Select the extension folder

3. Open a supported problem page.

4. Click the extension icon.

5. In the popup:

• configure the backend and AI provider if needed
• click Solve Current Page
• review the output in the popup
• use Copy Code or Copy Notes

How To Use

1. Double-click run_backend.cmd.
2. In Chrome, open chrome://extensions.
3. Turn on Developer mode.
4. Click Load unpacked and choose the extension folder inside this project.
5. Open a supported problem page.
6. Click Solve Current Page in the popup.
7. Copy the generated code manually into the site editor if you want to test it.

Compliance Notes

• Keep submissions manual.
• Use it as a visible drafting tool, not a hidden automation layer.
• Avoid excessive request loops or unattended interaction on third-party platforms.
• If a site changes its DOM or editor integration, update selectors instead of trying to bypass platform behavior.

Backend Configuration

By default the backend can run in three useful modes:

• Local Fallback: free, offline, weaker starter drafts
• OpenAI Compatible: any hosted OpenAI-compatible API
• Ollama Local: free local model runtime on your machine

The popup now includes a Quick Setup selector for the common paths.

Fastest free setup: Ollama

1. Install Ollama.
2. Pull a local model, for example:

ollama pull llama3.2

3. Keep Ollama running.
4. In the popup choose:

• Quick Setup: Ollama Local (Free)
• Model: llama3.2 or any model you have pulled
• API URL: http://127.0.0.1:11434/v1/chat/completions
• API Key: leave blank

You can also copy provider_config.ollama.example.json to backend/provider_config.local.json.

OpenAI or another hosted provider

The backend now supports both of these OpenAI-compatible endpoint styles:

• https://.../v1/responses
• https://.../v1/chat/completions

That means you can use OpenAI-compatible providers beyond the OpenAI API itself as long as you supply:

• a supported model name
• the provider's base URL for responses or chat/completions
• an API key when the provider requires one

The backend also supports environment variables before startup:

• LEETBOT_PROVIDER=openai
• LEETBOT_API_KEY=...
• LEETBOT_MODEL=...
• LEETBOT_API_URL=...

The backend uses only the Python standard library, so no package install is required.

Verification

Run:

run_backend_tests.cmd

This checks the provider fallback logic and the HTTP API round-trip.

Notes

• DOM selectors can change over time, so the content script is structured around per-platform adapters for easier updates.
• The extension now uses a popup-only workflow to avoid interfering with coding site editors or navigation.

Open Source

This project is available under the MIT License.

Before publishing to GitHub:

• do not commit backend/provider_config.local.json
• rotate any API key that has already been pasted into chat, screenshots, or local files you may upload
• keep only the example config files in the public repo