⚡ promptforge

turn rough ideas into production-grade prompts. free. open source.

most people fail at prompting because they skip structure, context, constraints, and examples. promptforge fills that gap automatically.

describe what you want to build → get a perfectly structured prompt → paste into any LLM → ship.

how it works

"i want a crypto portfolio tracker"

    ↓ promptforge asks smart follow-up questions

    ↓ generates a 2,000+ token production-grade prompt with:
      • expert role assignment
      • specific tech stack
      • complete file structure
      • design specs (hex colors, fonts, layout)
      • numbered requirements
      • constraints to prevent mistakes
      • step-by-step build order

    ↓ copy → paste into claude / chatgpt / gemini / qwen

    ↓ LLM builds exactly what you described. first try.

features

• adaptive interview — asks smart follow-up questions based on your idea's complexity
• production-grade output — prompts follow anthropic's best practices (xml tags, role assignment, chain of thought)
• works with any LLM — claude, chatgpt, gemini, qwen, local models
• 12+ templates — websites, bots, apis, agents, scripts, apps
• quality scoring — shows why your prompt works and what's included
• 100% free — powered by free LLM APIs (groq, qwen, gemini, mistral)
• no signup required — just type and forge
• open source — MIT license, contribute templates easily

quick start

# clone the repo
git clone https://github.com/Vt01nft/promptforge.git
cd promptforge

# install dependencies
npm install

# copy env file and add your free API keys
cp .env.example .env.local
# edit .env.local — add at least one key (groq recommended)

# run locally
npm run dev

open http://localhost:3000

free API keys

all free, no credit card required:

provider	free tier	get key
groq (recommended)	14,400 req/day	console.groq.com
qwen 3.5	free quota	dashscope.console.aliyun.com
google gemini	500 req/day	aistudio.google.com
mistral	1B tokens/month	console.mistral.ai
openrouter	50 req/day	openrouter.ai

add at least one key. the app auto-cascades through available providers.

tech stack

• next.js 15 — app router, typescript, server components
• tailwind css — dark theme, custom design system
• framer motion — smooth animations
• zustand — lightweight state management
• openai sdk — unified LLM interface (works with all providers)
• vercel — free deployment

project structure

promptforge/
├── app/
│   ├── page.tsx                # landing page
│   ├── forge/page.tsx          # main forge interface
│   ├── templates/page.tsx      # template library
│   └── api/
│       ├── generate/route.ts   # prompt generation endpoint
│       └── interview/route.ts  # follow-up questions endpoint
├── components/
│   ├── forge/                  # forge UI components
│   ├── shared/                 # navbar, shared components
│   └── landing/                # landing page sections
├── lib/
│   ├── llm/
│   │   ├── providers.ts        # LLM provider configs
│   │   ├── cascade.ts          # auto-fallback logic
│   │   └── meta-prompt.ts      # the meta-prompting engine
│   ├── templates/              # prompt templates
│   ├── store.ts                # zustand state
│   └── utils.ts                # utilities
└── public/

contributing

adding a template is the easiest way to contribute:

1. fork the repo
2. add your template to lib/templates/index.ts
3. submit a PR

templates are just objects with an id, name, category, description, and starter idea. see existing templates for the format.

other ways to contribute:

• improve the meta-prompting engine
• add new LLM providers
• improve the UI/UX
• fix bugs
• add tests

deploy your own

1. click the button above
2. add your API keys in vercel environment settings
3. deploy — live in 60 seconds

license

MIT — do whatever you want with it.

---

star this repo if it helped you ship faster ⭐