StackSecure AI

A security-themed dashboard that fingerprints a website's technology stack and cross-checks it against real vulnerability data.

Built on a React admin template, with a working analyzer bolted on:

• Stack Analyzer (functional) — give it a URL; an Express backend fetches the page, identifies frameworks and libraries from the HTML and response headers (TECH_SIGNATURES — React, Next.js, Vue, Angular, jQuery, WordPress, Laravel, Django, Express, Rails, …), then queries OSV.dev for known advisories on the detected packages and versions.
• Hunter's Tips (optional) — a HuggingFace Mistral-7B-Instruct call turns a finding into specific testing guidance. Needs HF_KEY_* env vars; skipped if unset.
• Global Threat Feed and Vulnerability Registry (static) — curated datasets in src/data/*.ts (threats.ts, vulnerabilities.ts, stackTechs.ts) used to populate the UI. These are illustrative, not a live feed.

Stack

• Frontend — React 19, Vite, TypeScript, Tailwind, Framer Motion (admin template base)
• Backend — Express (server/index.js), no database; an in-memory cache
• External — OSV.dev API, HuggingFace Inference API
• Deploy — Vercel (vercel.json builds the SPA and runs server/ as a serverless function)

Run locally

npm install

# backend on :4000
node server/dev.js        # or: node server/index.js

# frontend on :5173
npm run dev

The frontend calls http://localhost:4000/api by default (see src/pages/stack-analyzer/AnalysisOutput.tsx). Copy .env.example to .env and add HuggingFace tokens if you want the AI tips.

Layout

src/
  pages/
    dashboard/         overview + live detect widget
    stack-analyzer/    URL → tech fingerprint → OSV lookup
    vulnerabilities/   registry view (static data)
    authentication/    template auth screens
    security-settings/
  data/                curated threat / vuln / tech datasets
server/
  index.js             /detect, /analyze, /health + OSV integration
  aiService.js          HuggingFace Mistral wrapper (token rollover)

Honest scope

This started from a paid dashboard template, so a lot of the surrounding UI (mentors, tasks, avatars) is template scaffolding. The parts that are mine and actually work are the tech-fingerprint backend, the OSV integration, and the analyzer flow.