fitcheck

Compare a resume with a job description and get a fit score, requirement evidence, missing keywords and factual rewrite suggestions. The interface supports English and Portuguese. This project started as a personal agent used day to day and became a product.

Run locally

Use Node.js 22 and npm.

npm install
cp .env.example .env.local
npm run dev

Set LLM_API_KEY, LLM_BASE_URL and LLM_MODEL in .env.local before analyzing. The base URL must expose an OpenAI-compatible chat completions API. Optionally set APP_PASSWORD to require a password. Open http://localhost:3000 and select Load example to use fictional documents.

Technical decisions

Next.js App Router, TypeScript, Tailwind CSS and lucide-react provide the interface. The OpenAI SDK calls the configured gateway only on the server. Zod validates structured results. A plain bilingual object supplies interface text, and localStorage stores only the language choice. Resume text and reports are not persisted by this app; the gateway's own retention policy still applies.

npm run typecheck
npm run build
npm start

Validation status

Build and type checking passed during initial setup. Live model analysis and browser visual checks are still pending.

Demo

Next steps

Future versions may add PDF upload, saved profiles, accounts, analysis history, per-IP rate limiting, full resume generation, a database and automated tests.