How to vibe codeResume Worded

Prerequisites

• Node.jsfree

  Required runtime for running Next.js and backend document parsing scripts

• GitHubfree

  Source control and deployment pipeline to Vercel

• Anthropic API KeyPay-as-you-go (~$10 starting credit)

  Powers resume evaluation heuristics and AutoFix rewrites

AI coding tools

• Claude Code

  Best-in-class coding agent for scaffolding multi-file Next.js apps, setting up document parsers, and handling complex prompt engineering loops

• Cursor

  Ideal for fine-tuning UI components, CSS layout adjustments, and reviewing prompt diffs interactively

Build guide

1. 01Scaffold Next.js App & UI Shell

   Initialize the Next.js project with Tailwind CSS, shadcn/ui components, and setup local routing for dashboard, resume scanner, and targeted keyword tools.

   Create a new Next.js application using App Router, TypeScript, and Tailwind CSS. Install shadcn/ui primitives including buttons, cards, progress bars, tabs, and dialogs. Build a clean, professional SaaS dashboard layout styled with a slate and indigo color palette, featuring a sidebar navigation for 'Score My Resume', 'AutoFix', 'Smart Target', and 'Cover Letter Generator'. Ensure the UI is fully responsive and clean.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

2. 02Document Parsing Engine (PDF & DOCX)

   Implement server-side file upload ingestion using `pdf-parse` for PDFs and `mammoth` for DOCX files, extracting raw text and structuring sections.

   Implement a Next.js API route / server action that accepts PDF and DOCX file uploads (max 2MB) using standard multipart form data. Use `pdf-parse` for parsing PDF files and `mammoth` for parsing .docx files into clean text. Write a robust text extraction utility that normalizes whitespace and attempts to segment the raw text into standard resume sections: Contact Info, Summary, Experience, Education, and Skills. Handle malformed files gracefully and return structured JSON containing section strings and raw text.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

3. 03Resume Scoring & Heuristic Rules Engine

   Build the evaluation engine using the Vercel AI SDK and Anthropic API to run 30+ recruiter checks, outputting a score out of 100 and line-by-line critiques.

   Build an evaluation service using the Vercel AI SDK with Anthropic Sonnet. The service takes parsed resume text and runs a structured evaluation against 30+ recruiter criteria (e.g. quantified impact metrics in bullets, active verbs, correct section structure, contact details present, length check). Use structured JSON output with Zod validation to return a numerical score out of 100, an array of passed checks, and an array of specific line flags with explanations and line numbers. Display these results in a clean breakdown UI with progress bars and color-coded flags.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

4. 04AutoFix Bullet Point Rewriter

   Create the AutoFix interactive agent that surfaces weak bullet points, prompts the user for context, and generates rewritten impact-driven bullets.

   Build an 'AutoFix' feature component in Next.js. It takes flagged weak bullet points from the resume scanner, sends them to an LLM endpoint with prompt instructions to rewrite them using strong action verbs and quantified impact (asking clarifying questions where metrics are missing), and presents the original vs. rewritten bullets side-by-side. Include interactive checkboxes allowing the user to approve or reject each individual line rewrite. Add an export button that generates and downloads an updated .docx file containing the approved changes.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

5. 05Smart Target Keyword Matching

   Implement job description comparison to calculate keyword match percentages, showing matched (green) and missing (amber) keywords.

   Implement a 'Smart Target' feature where users paste a job description alongside their uploaded resume. Build an extraction and matching utility using LLM tool calls or semantic analysis that extracts key required skills, tools, and qualifications from the job description and checks them against the resume text. Output a relevancy score (0-100) and two categorized lists: matched keywords (green check) and missing keywords (amber warning). Allow users to click missing keywords to see suggested bullet insertions.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

6. 06Database Persistence & Polish

   Connect Turso SQLite database using Drizzle ORM to save resume versions, scan history, and user sessions via better-auth.

   Configure Turso SQLite with Drizzle ORM to persist user accounts (via better-auth), uploaded resume metadata, raw text, scan scores, and AutoFix revision history. Create a history dashboard page where users can view past resume uploads, track their score progression over time (e.g. 42 -> 61 -> 93), and reload previous review sessions. Add error boundaries, toast notifications for uploads, and final polish across all tool views.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

Cost vs paying for Resume Worded