https://github.com/rohit-pakhre09/codesage-ai
CodeSage is an AI-powered code review platform that analyzes code quality, identifies issues, and provides clear, explainable improvements using modern AI models.
https://github.com/rohit-pakhre09/codesage-ai
code-reviewer gen-ai groq-api
Last synced: 2 months ago
JSON representation
CodeSage is an AI-powered code review platform that analyzes code quality, identifies issues, and provides clear, explainable improvements using modern AI models.
- Host: GitHub
- URL: https://github.com/rohit-pakhre09/codesage-ai
- Owner: Rohit-Pakhre09
- License: mit
- Created: 2026-01-04T07:40:44.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2026-04-19T07:42:18.000Z (3 months ago)
- Last Synced: 2026-04-19T09:33:29.474Z (3 months ago)
- Topics: code-reviewer, gen-ai, groq-api
- Language: TypeScript
- Homepage: https://codesage-ai-review.vercel.app/
- Size: 447 KB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
CodeSage AI | AI Powered Code Reviewer
Paste code. Get clarity.
Structured reviews, optimized code, and explanation depth that adapts to the developer reading it.
Overview |
Sneak Peek |
Why It Feels Different |
Experience Flow |
Quick Start |
Project Structure
---
## Overview
CodeSage AI is a modern code review experience built with Next.js. It is designed for developers who want feedback that is fast, readable, and actionable instead of noisy or vague.
Drop in a snippet, choose your language, pick the explanation level, and CodeSage AI returns:
- a summary of code quality
- detected issues with severity
- improvements worth making
- an optimized version of the submitted code
- a clear explanation matched to the chosen level
> The goal is simple: make code review feel less like a wall of text and more like a guided workspace.
## Sneak Peek
## Why It Feels Different
Structured Output
The result is organized into clear sections instead of a single messy response block.
Level-Based Guidance
Beginner, intermediate, and advanced modes make the same review readable for different developers.
Safer Reliability
If the AI response is weak or incomplete, the backend repairs it or falls back locally.
## Experience Flow
```text
Landing Page -> Review Workspace -> AI Review -> Repair Pass if Needed -> Clean Result
\
-> Local Fallback if Provider Fails
```
### What users do
1. Open the intro page.
2. Enter the review workspace.
3. Paste a code snippet into the Monaco editor.
4. Select a language and explanation level.
5. Run the review.
6. Inspect issues, improvements, optimized code, and explanation.
7. Copy the optimized code directly from its section.
## Feature Highlights
### Smart Review Layout
- dedicated intro page on `/`
- focused review workspace on `/review`
- organized result cards for quick scanning
- cleaner UI for summary, issues, improvements, and explanation
### Review Engine
- Groq-powered structured code review
- JSON-shaped responses
- explanation completeness checks
- optimized code repair logic for common AI misses
- local fallback review when provider output is unavailable
### Developer Experience
- Monaco editor integration
- language selection
- explanation depth selection
- copy action for optimized code only
- resilient API route for incomplete AI output
## Modes
| Level | Style |
| --- | --- |
| `beginner` | Foundational guidance and simpler explanations |
| `intermediate` | Balanced practical feedback |
| `advanced` | Sharper technical detail and engineering-focused explanation |
## Example Review Shape
```json
{
"summary": "The function has a runtime bug caused by an undefined variable.",
"score": 5,
"issues": [
{
"line": 1,
"message": "Undefined variable 'n'. Expected 'b' but got 'n'.",
"severity": "high"
}
],
"improvements": [
"Fix the incorrect variable reference.",
"Add simple tests for expected inputs."
],
"optimizedCode": "function add(a, b) {\n return a + b;\n}",
"explanation": "The updated function replaces the undefined variable with the intended parameter so the code behaves correctly."
}
```
## Quick Start
### 1. Install dependencies
```bash
npm install
```
### 2. Add environment variables
Create `.env.local` in the project root:
```env
GROQ_API_KEY=your_groq_api_key_here
GROQ_MODEL=llama-3.1-8b-instant
```
### 3. Start the app
```bash
npm run dev
```
Open `http://localhost:3000`.
## Scripts
```bash
npm run dev
npm run build
npm run start
npm run lint
npm run format
```
## Tech Stack
| Layer | Technology |
| --- | --- |
| Framework | Next.js 16 |
| UI | React 19 |
| Language | TypeScript |
| Styling | Tailwind CSS v4 |
| Editor | Monaco Editor |
| Validation | Zod |
| AI Provider | Groq API |
## Project Structure
```text
src/
app/
page.tsx
review/
page.tsx
api/
review/
route.ts
components/
editor/
CodeEditor.tsx
lib/
ai/
aiClient.ts
fallbackReview.ts
promptBuilder.ts
validators/
reviewResponse.schema.ts
public/
CodeSage-AI.png
codesage-mark.svg
```
## Current Strengths
- polished intro page before entering the tool
- dedicated review workspace
- structured and readable review results
- optimized-code copy action inside the relevant section
- improved repair path for incomplete AI outputs
- local fallback behavior that keeps the app usable
## Notes
- Groq is the current primary provider.
- When the AI response is partial, CodeSage AI attempts to repair weak fields before falling back.
- The fallback layer is designed to protect the UX instead of failing hard.
## License
This project is licensed under the [MIT License](./LICENSE).