14 LANGUAGES · CLI + MCP

Your Codebase

Has DNA.

Extract It.

A static analysis pipeline that gives any LLM instant architectural understanding — zero inference tokens wasted. Analyze once. Give every session full codebase context.

$npx @qlucent/code-dna analyze

Star on GitHub npmView Package

code-dna analyzeLIVE

▋

4,821

OUTPUT TOKENS

~12k

VS RAW

1.2s

TIME

The Problem

Every session burns a token tax just to establish context.

“A model is not the right tool to read files. A static analysis pipeline is.”

~12k

tokens / session wasted

on a median TypeScript monorepo

WITHOUT CODE-DNA

Reading every file from scratch95%

Re-discovering patterns each session88%

Guessing architecture context76%

Missing conventions & hotspots82%

WITH CODE-DNA

Complete architecture mappre-analyzed

Conventions extracted oncepre-analyzed

Risk surface identifiedpre-analyzed

Full dependency graphpre-analyzed

~95%

context consumed

~8%

with code-dna

Analyzed once, reused every session. Your LLM starts informed.

The Pipeline

Four layers. One output.

Like DNA sequencing, each layer extracts different signal from the same source.

01A—TStructural Skeleton

02G—CGit Archaeology

03T—APattern Inference

04C—GRisk Surface

WHAT EXISTS

Structural Skeleton

File tree, module map, dependency graph, symbol extraction across all entry points. Every import resolved, every export catalogued.

File tree with depth control

Module dependency graph

Entry point detection

Symbol extraction

A—T

SCROLL TO EXPLORE LAYERS · 1/4

Features

Built for the modern AI workflow.

Every feature designed to fit seamlessly into how you already work — whether you are in the terminal, an AI editor, or building on top of the API.

Languages Supported

TypeScript, JavaScript, Python, Go, Rust, Java, C#, Kotlin, Swift, Ruby, PHP, C/C++, Solidity, Vue. Framework-aware detection for every major ecosystem.

TSJSPyGoRustJavaC#KotlinSwiftRubyPHPC/C++SolVue

MCP Server

Native integration with Claude Code, Cursor, and Cline. Fresh DNA on every query.

Claude CodeCursorCline

Watch Mode

File system monitoring with smart debounce. DNA stays current as you write code.

LiveDebounced

DNA Diff

Track structural changes between analyses. See exactly what evolved in your architecture.

VersionedDiffable

Token Budget

Adaptive output sizing. Set a ceiling — 5K or 50K tokens — for any context window.

5K25K50K

Zero Infrastructure

No databases. No servers. No auth. Pure static analysis — runs anywhere, offline.

OfflineFastPortable

Markdown + YAML

Two output formats. Human-readable Markdown for LLMs, structured YAML for pipelines.

.md.yaml

Output

See the DNA report.

Actual output from a real codebase analysis. Deterministic, structured, LLM-optimized. Paste directly into any context window.

4,821

TOTAL TOKENS

vs ~12k raw

1.2s

ANALYSIS TIME

247 files

LAYERS

structural · git · pattern · risk

CODEBASE-DNA.md

generated

# CODEBASE-DNA.md

# Generated by code-dna

# Analyzed: 2026-03-26T09:15:22Z

# Files: 247 | Languages: 3 | Tokens: 4,821

## GENOME

project:           code-dna

version:           latest

primary_language:  TypeScript (91.4%)

architecture:      Pipeline + CLI (layered)

framework:         None (Node.js runtime)

entry_points:      [src/cli/index.ts, src/mcp/server.ts]

## MODULE MAP

src/

  analyzers/       [4 layers, 14 analyzers]

  parsers/         [14 language parsers]

  cli/             [analyze, watch, diff, mcp]

  mcp/             [MCP server + tools]

## TOP CONVENTIONS

naming:    camelCase functions, PascalCase classes

testing:   Vitest, co-located __tests__

imports:   ESM, path aliases (@/)

commits:   Conventional Commits enforced

## RISK SURFACE

HIGH:  src/analyzers/pipeline.ts    centrality: 0.94, churn: 22

MED:   src/parsers/index.ts         churn: high, test-coverage: 40%

LOW:   src/cli/commands.ts          stable, tested, low-centrality

14 Languages · One Pipeline

TypeScript

JavaScript

Python

Rust

Java

Kotlin

Swift

Ruby

PHP

C/C++

Solidity

Vue

TypeScript

JavaScript

Python

Rust

Java

Kotlin

Swift

Ruby

PHP

C/C++

Solidity

Vue

Integration

Plug into your workflow.

Three integration paths. All take under a minute. Once configured, code-dna runs in the background and keeps your LLM context current.

.claude/mcp.json

{
  "mcpServers": {
    "code-dna": {
      "command": "npx",
      "args": [
        "@qlucent/code-dna",
        "mcp",
        "--watch"
      ]
    }
  }
}

Paste into your Claude Code MCP settings. DNA refreshes automatically on file changes.

Get Started

Decode your
codebase.

Three steps. Under two minutes. No server, no database, no configuration file. One command gives your LLM a complete architectural picture of any codebase.

npx @qlucent/code-dna analyze

Or install globally: npm i -g @qlucent/code-dna

STAR ON GITHUB npmVIEW PACKAGE

Quick start — three steps

Step 01

Install

npm i -g @qlucent/code-dna

Global install for repeated use across projects. Or run via npx — no install required.

Step 02

Analyze

code-dna analyze ./my-project

Generates CODEBASE-DNA.md in seconds. Outputs a compact 4-layer DNA report — no config required.

Step 03

Integrate

→ MCP config in any AI editor

Drop into Claude Code, Cursor, or any MCP-compatible editor. DNA refreshes automatically on changes.