⚡️ C O D E - X - R A Y

The High-Performance AST Scanner for the Modern Web

Give your AI Agents "Eyes" into your codebase. Scan 10,000 files in seconds. Extract symbols instantly. Zero overhead.

Report Bug · Request Feature

🔮 The Problem

Node.js is fantastic, but it struggles with heavy computation. Trying to parse thousands of TypeScript or Python files in a single thread is slow, fragile, and memory-intensive.

⚡️ The Solution: Code X-Ray

We moved the heavy lifting to Rust. By bridging Node.js with a compiled Rust binary, we bypass the event loop entirely, utilizing Parallel Computing to scan your project at the speed of disk I/O.

🔥 Why Developers Choose X-Ray

| 🚀 Blazing Performance | 🛡️ Bulletproof Safety | 🧠 Intelligent Parsing | | :--- | :--- | :--- | | Uses Rayon to multithread across all CPU cores. Scans ~1.5ms per file. | Sandboxed file access. Respects .gitignore automatically. No crashes. | Powered by Tree-Sitter. Understands code structure, not just regex matches. |

📦 Installation

Add it to your project with a single command. It detects your OS (Windows/Linux/Mac) and downloads the correct optimized binary automatically.

npm install code-x-ray

💻 Developer Experience We designed Code X-Ray to feel like a native part of your toolchain. It’s fully typed, asynchronous-ready, and zero-config.

1. The "Hello World" Scan Get a complete map of your project in 3 lines of code.

TypeScript

import { scanProject } from 'code-x-ray';

🚀 Fire up the engine (scans recursively)

console.time("✨ Magic Time");
const context = scanProject("./src");
console.timeEnd("✨ Magic Time");

console.log(`\n📦 Scanned ${context.files_scanned} files in ${context.duration_ms}ms`);

2. Powerful Filtering

Don't just scan—understand. Filter the raw AST data to find exactly what you need.

TypeScript

// Example: Find all 'TODO' comments or specific function definitions
const context = scanProject("./src");

// 🔍 Filter for TypeScript functions only
const functions = context.files
  .filter(f => f.language === 'TypeScript')
  .flatMap(f => f.symbols)
  .filter(s => s.kind === 'function');

console.table(functions.map(fn => ({
  Name: fn.name,
  Location: `L${fn.start.row}:${fn.start.column}`,
  Signature: fn.signature || '(unknown)'
})));

3. The "Context" Payload

The engine returns a clean, highly-structured JSON object optimized for LLM Context Windows (GPT-4, Claude, Llama 3).

JSON

{
  "root_dir": "./src",
  "stats": {
    "duration_ms": 142.5,
    "files_processed": 45,
    "threads_active": 12
  },
  "files": [
    {
      "path": "src/services/auth.ts",
      "language": "TypeScript",
      "size": 2048,
      "symbols": [
        {
          "name": "authenticateUser",
          "kind": "function",
          "range": { 
            "start": { "row": 15, "col": 0 }, 
            "end": { "row": 25, "col": 1 } 
          },
          "signature": "async (token: string) => Promise<User>"
        }
      ]
    }
  ]
}