RepoIntel

AI-powered repository intelligence for public GitHub repositories.

RepoIntel clones and structurally analyzes a repository, selects the most relevant source evidence, and uses task-routed Gemini models to produce an evidence-based engineering review covering architecture, security, code quality, testing, and production readiness.

What It Does

GitHub Repository
    ↓
Deterministic Structural Scan
    ↓
Evidence Selection
    ↓
Fast Repository Classification
    ↓
Deep Engineering Review
    ↓
Validated Intelligence Report

Current AI Pipeline

• Structural analysis: Deterministic Python (LangGraph workflow)
• Fast classification: Gemini 3.1 Flash Lite
• Primary engineering review: Gemini 3.5 Flash
• Report rendering: Deterministic Python
• Failure mode: Structural analysis fallback

The architecture uses task-based model routing rather than a supervisor loop, applying a strong, capable model (Gemini 3.5 Flash) only when deep semantic understanding is necessary, and utilizing a faster, more economical model (Gemini 3.1 Flash Lite) for simple classification tasks. An automatic model cascade is in place: if the primary model hits rate limits or is unavailable, requests failover seamlessly to a fallback model.

Features

• Evidence Selection: Deterministically selects the most critical files (Dockerfiles, main APIs, config) to stay within budget constraints.
• Model Cascade: Fails gracefully if Gemini 3.5 Flash is busy, switching to Gemini 3.1 Flash Lite automatically.
• Structural Fallback: If all AI models are unavailable (e.g., API keys revoked or quota exhausted), RepoIntel still generates a structural report detailing file counts, languages, CI presence, and basic health metrics.
• Real-Time Progress: Powered by FastAPI and WebSockets, users receive real-time granular progress (including per-stage timing) in the frontend.
• Safe Resource Limits: Shallow cloning restricts the download depth of repositories, and temporary repositories are safely cleaned up automatically.

Verified Run

RepoIntel successfully analyzed its own repository (V2 Runtime):

• 39 files scanned
• 2,595 lines of code
• 15 evidence files selected
• 31,354 evidence characters
• 2 successful AI calls
• Analysis mode: AI Enhanced
• Final engineering score: 6.3/10

Quick Start

1. Requirements

• Python 3.12+
• Gemini API Key

2. Setup

python -m venv venv
# Windows
.\venv\Scripts\Activate.ps1
# Linux/Mac
source venv/bin/activate

pip install -r requirements.txt

3. Configuration

Create a .env file in the root directory:

GEMINI_API_KEY=your_gemini_api_key_here

4. Run the Server

uvicorn src.main:app --reload --host 127.0.0.1 --port 8000

Open http://127.0.0.1:8000 in your browser.

Docker Deployment

You can also run RepoIntel using Docker Compose:

docker-compose up --build