#RepoLogs

AI-powered code quality analysis for GitHub repositories — right in your browser.

RepoLogs is a Chrome extension that injects an analysis button into any public GitHub repository page. One click triggers an intelligent repository-mapping and file-sampling pipeline, sends the code plus repository structure context to Google Gemini, and renders a detailed quality report — score, grade, architecture rating, security flags, strengths, weaknesses, and actionable recommendations — without leaving GitHub.

---

#Features

• One-click analysis — button injected directly into GitHub's repo header
• AI scoring — 0–100 score with letter grade (A–F) powered by Google Gemini
• Comprehensive report — summary, strengths, weaknesses, inconsistencies, recommendations (high / medium / low priority), detected tech stack, and security flags
• Architecture evaluation — dedicated rating with qualitative notes
• Intelligent file sampling — priority files (README, package.json, Dockerfile, CI configs) + centrality-based code file selection; ignores node_modules, dist, tests, lock files, and binaries
  • Dependency graph — all sampled files are scanned for import/require statements to build a directed dependency graph; each edge represents one file importing another. Files are then ranked by in-degree (number of other files that import them), so the most architecturally central modules rise to the top of the selection pool. Priority files always occupy the first slots regardless of in-degree.
• Repository structure context — the full repository file map is compacted into a filtered directory tree and included in the Gemini prompt so the model knows which files exist even when only a subset of file contents is loaded
• Deep mode — increases per-file line budget from 150 → 350 lines for larger codebases
• Multiple Gemini models — Gemini 2.5 Flash (default), 2.5 Flash Lite, or 2.5 Pro
• Free tier — one free analysis using the system key; unlimited with your own Gemini API key
• Result caching — analyses are cached by commit SHA for instant re-viewing; individual file contents are also cached locally so repeat analyses skip redundant network requests
• Dark / light theme — follows the OS preference

---

#Screenshots

---

#How It Works

User clicks "RepoLogs" on GitHub
        │
        ▼
Content Script → Background Worker
        │
        ├─ 1. Fetch repo metadata (default branch, latest SHA)
        ├─ 2. Fetch full file tree via GitHub API
        ├─ 3. Build a filtered directory tree from the full repo file list
        ├─ 4. Sample up to 80 files (priority + code files)
        ├─ 5. Read file contents — 8 concurrent requests via raw CDN (no API rate limit), max 150/350 lines each
        ├─ 6. Build dependency graph → rank files by import centrality
        ├─ 7. Select top 40 files within token budget
        ├─ 8. Call Gemini API with sampled code + repository structure prompt context
        └─ 9. Cache result and send to content script
                │
                ▼
        Modal rendered with full report

---

#Tech Stack

---

#Installation (Development)

#Prerequisites

• Node.js ≥ 18
• npm ≥ 9
• A Google AI Studio account to generate a Gemini API key

#Steps

# 1. Clone the repository
git clone https://github.com/your-org/RepoLogs_GithubExtension.git
cd RepoLogs_GithubExtension

# 2. Install dependencies
npm install

# 3. Configure environment variables
cp .env.example .env
# Edit .env and set VITE_GEMINI_SYSTEM_KEY to your Gemini API key (optional — enables free tier)

# 4. Build for production
npm run build

# 5. Load the extension in Chrome
#    a. Open chrome://extensions/
#    b. Enable "Developer mode" (top-right toggle)
#    c. Click "Load unpacked"
#    d. Select the dist/ folder

#Development (hot reload)

npm run dev
# Vite dev server starts; reload the extension in chrome://extensions/ after the first build

---

#Available Scripts

---

#Environment Variables

Create a .env file at the project root (see .env.example):

---

#Extension Settings

All settings are persisted in Chrome's local storage.

Settings are accessible via the extension popup (click the toolbar icon).

---

#Project Structure

src/
├── manifest.ts              # Chrome extension manifest (Manifest V3)
├── background/
│   └── worker.ts            # Service worker — analysis orchestration
├── content/
│   ├── index.ts             # Content script entry point
│   ├── button.ts            # Injects "RepoLogs" button into GitHub pages
│   └── modal.ts             # Result modal UI (score ring, report sections)
├── popup/
│   ├── index.ts             # Popup controller
│   ├── index.html           # Popup markup
│   └── popup.css            # Popup styles
└── shared/
    ├── types.ts             # Shared TypeScript interfaces
    ├── api-key-manager.ts   # System key vs. user key resolution
    ├── gemini.ts            # Gemini API client and analysis request assembly
    ├── github.ts            # GitHub API client (tree, file content)
    ├── sampler.ts           # Intelligent file sampling, centrality ranking, and directory-tree compaction
    ├── prompt.ts            # Gemini system + user prompts, including repository structure context
    └── storage.ts           # Chrome storage wrapper + cache management

---

#Analysis Report Structure

The Gemini response is parsed into the following structure:

interface AnalysisResult {
  score: number;           // 0–100
  grade: string;           // A | B | C | D | F
  summary: string;
  strengths: string[];
  weaknesses: string[];
  inconsistencies: string[];
  recommendations: Recommendation[];  // { text, priority: 'high'|'medium'|'low' }
  architecture: { rating: string; notes: string };
  techStack: string[];
  securityFlags: string[];
}

---

#Chrome Permissions

---

#Contributing

1. Fork the repository and create a feature branch
2. Run npm run dev and load the dist/ folder as an unpacked extension
3. Make your changes — the service worker auto-reloads; content scripts require a manual extension reload in chrome://extensions/
4. Open a pull request with a clear description of the change

---

#License

MIT — see LICENSE for details.

---

Analysis is generated by AI. Use as a reference, not as an absolute truth.