resumeforge

Build ATS-friendly resumes in minutes - 100% local, no backend, no account.

resumeforge is an open-source, AI-assisted resume builder that runs entirely in your browser. Create, customize, import, and export professional, ATS-friendly resumes with complete control of your data. There is no backend, no account, and no telemetry - your resume is stored locally (IndexedDB, with a localStorage fallback) and never leaves your device. The optional AI assistant uses your own API key and talks directly to the provider from your browser.

---

✨ Features

Builder

• 11 resume sections - Professional summary, Experience, Education, Projects, Skills, Certifications, Achievements, Publications, Languages, Interests, and References.
• Add, remove, reorder, and duplicate sections and entries with drag & drop (powered by dnd-kit).
• Live preview that updates as you type.
• Autosave + draft recovery, undo/redo, and a clean form built on react-hook-form + zod validation.

Templates

• 9 ATS-aware templates with instant switching: ATS Pro (most used), Minimal ATS, Software Engineer, Modern Professional, Harvard, Startup Founder, Executive, Academic CV, and Creative.
• Customization per resume: accent color, font, spacing, section order, and page size (A4 / Letter).
• Single-column templates are flagged ATS-safe so you always know what a parser sees.

Scoring

• Real-time ATS scoring & completeness as you edit.
• 9 ATS checks - name & title, contact details, summary, experience, bullet points, quantified impact, skills, education, and ATS-safe layout.
• 5 category scores - ATS readability, Keywords, Formatting, Experience quality, and Skills coverage - plus prioritized, actionable suggestions.

Import & export

• JSON import to bring an existing resume in.
• A privacy-first "convert your old PDF/DOCX" flow: paste your old resume into ChatGPT (or any LLM) with the provided prompt, get clean JSON back, and import it - your document never touches a resumeforge server because there isn't one.
• PDF export for A4 and Letter: print to a PDF with selectable text, or download a rasterized PDF (jsPDF + html2canvas-pro).

AI assistant (optional, bring-your-own-key)

• Opt-in assistant that runs client-side with your own API key.
• Supports OpenAI, Anthropic, Google Gemini, DeepSeek, and Groq.
• Requests go directly from your browser to the provider - the key is stored locally and never sent anywhere else.

Everything else

• Dark mode, keyboard shortcuts (undo / redo / save), and full keyboard & ARIA accessibility built on Radix primitives.
• 100% local storage - IndexedDB with a transparent localStorage fallback.

---

📸 Screenshots

View	Preview
Landing
Dashboard
Template gallery
Import

---

🧰 Tech stack

Area	Technology
Framework	Next.js 16 (App Router, Turbopack)
UI library	React 19
Language	TypeScript 5 (strict)
Styling	Tailwind CSS v4
Components	shadcn-style UI on Radix UI primitives
State	Zustand 5
Forms	react-hook-form 7 + zod 4
Drag & drop	dnd-kit (core, sortable, utilities)
PDF export	jsPDF 4 + html2canvas-pro 2
Icons	lucide-react
Notifications	sonner
Theming	next-themes
Storage	IndexedDB + localStorage (custom abstraction)
Package manager	Bun
Linting	ESLint 9 (eslint-config-next)

---

🏗️ Architecture

resumeforge uses a feature-based structure. Application code lives in src/ and is split into three top-level zones:

• src/app - Next.js App Router entry points (layout, routes, global styles). Route files stay thin and delegate to feature modules.
• src/features - self-contained domain modules. Each feature owns its store, hooks, services, types, and utilities, and features never import each other.
• src/global - shared building blocks used across features: UI components, config, constants, hooks, the storage abstraction, the scoring engine, and the template engine.

src/
├── app/                       # Next.js App Router (thin route files)
│   ├── layout.tsx
│   └── globals.css
├── features/                  # Domain modules (never import each other)
│   ├── ai/                    # Opt-in, bring-your-own-key AI assistant
│   │   ├── services/providers # Provider-abstracted client (OpenAI / Anthropic / …)
│   │   └── @types
│   ├── export/                # PDF export pipeline (print + rasterized)
│   │   └── @types
│   └── resumes/               # The resume editor + persistence
│       ├── hooks/             # useAutosave, useEditorShortcuts, useResumeList
│       ├── services/          # resume.repository.ts
│       ├── store/             # editor.store.ts (Zustand, undo/redo)
│       └── utils/
└── global/                    # Shared, cross-feature code
    ├── components/            # ui/ (Radix + shadcn-style) and shared/
    ├── config/                # site.ts (product identity)
    ├── constants/             # sections.ts (the section registry)
    ├── hooks/
    ├── lib/
    │   ├── scoring/           # ATS checks, category scores, suggestions
    │   └── storage/           # KeyValueStore abstraction (IndexedDB + localStorage)
    ├── templates/             # Template engine
    │   ├── components/        # Shared layouts (single / two column)
    │   ├── definitions/       # One file per template (TemplateDefinition)
    │   └── registry.ts        # Registers + looks up templates
    └── @types/

Key abstractions

• Storage abstraction (src/global/lib/storage) - everything goes through a single async KeyValueStore contract. The engine resolves to IndexedDB when available and transparently falls back to localStorage, and is SSR-safe (no-ops on the server), so callers never branch on the backend.
• Provider-abstracted AI (src/features/ai) - a static provider registry describes each provider's base URL, models, and wire format (apiShape: "openai" for OpenAI-compatible /chat/completions, or "anthropic" for the Messages API). Only the user's chosen provider id, model, and key are persisted (locally). All calls run browser → provider.
• Template engine (src/global/templates) - each template is a TemplateDefinition (id, name, category, columns, atsSafe flag, default theme, and a Renderer component). Definitions are registered in registry.ts; renderers reuse shared single-column and two-column layouts. A metadata-only projection (TemplateMeta) powers the gallery.

---

🚀 Getting started

Prerequisites

• Node.js 20.9+
• Bun (package manager and runtime)

Install & run

# 1. Clone
git clone https://github.com/ahmad2point0/resumeforge.git
cd resumeforge/web-app

# 2. Install dependencies
bun install

# 3. Start the dev server (Next.js dev runs Turbopack by default)
bun dev

Open http://localhost:3000 in your browser.

Scripts

Command	Description
bun dev	Start the development server (Turbopack).
bun run build	Create a production build.
bun run start	Serve the production build.
bun run lint	Run ESLint.

---

🔒 Privacy & data

resumeforge is local-first by design:

• No backend, no account, no telemetry. Your resume data lives only in your browser via IndexedDB (with a localStorage fallback).
• Export & import - back up or move a resume by exporting it as JSON and importing it on another device or browser. To migrate an existing PDF/DOCX, use the privacy-first flow: convert it to JSON with an LLM of your choice, then import the JSON.
• AI keys stay local. If you enable the optional AI assistant, your API key is stored on your device only, and requests go directly from your browser to the provider (OpenAI / Anthropic / Gemini / DeepSeek / Groq). resumeforge never sees your key or your prompts.

---

🗺️ Roadmap

See ROADMAP.md for the full plan. Highlights of what's planned next:

• Multi-page resume support and in-browser PDF text extraction for import.
• A cover-letter generator and a "tailor this resume to a job description" workflow.

---

🤝 Contributing

Contributions are welcome! Please read CONTRIBUTING.md for setup instructions, project conventions, and how to add a new template or section.

---

📄 License

Released under the MIT License.