Profoka: English Writing Tool and AI Humanizer

Profoka Humanizer, a personal English writing tool: side-by-side editor with goal, style, CEFR level (B1–C2/native), tone, readability and audience controls, chained rewrites, copy button, all powered by the built-in AI through a secure Express route. Visitors sign up free (no paywall) and confirm their email to start rewriting. You can own this platform today!

Preview and Live Test URL: https://profoka.com/

NB: Agent API Keys are currently disabled, live testing will not work effectively.

The platform is ready for ESL learners, content creators, and writers who want to adapt their own work to different contexts and audiences. Users can chain transformations together if they want to apply multiple changes.

Core Features:

• Paraphrasing – rewrite text for clarity, conciseness, or different audiences

• Style converter – switch between formal/casual and academic/conversational modes

• Proficiency levels – adapt text to B1, B2, C1, C2, or native English speaker level

• Tone adjuster – choose confident, cautious, friendly, or professional tones

• Readability control – simplify complex writing or elevate basic text

User Experience:

• Side-by-side comparison of original and transformed text

• Simple dropdown controls to select your desired transformation

• Copy-to-clipboard for easy use

• Free signup with email confirmation—no paywall

• Multi-agent pipeline: each selection (goal, style, CEFR level, tone, readability, audience) is its own agent with a dedicated system prompt, chained in sequence.

• All prompts live in one editable, documented file: apps/api/src/config/agent-prompts.js — edit a prompt, add options, restart; the UI reads the option list from GET /transform/options.

• Long texts (1200+ words) are split into ~4.5k-char paragraph-safe chunks per agent; limit raised to 40k chars, rate limiting kept, and the UI shows progress plus the applied agent chain and clear error feedback.

OpenAI Integration:

• API key configured in .env with a placeholder (you'll replace it with your own real key after purchasing this App) 

  

• Secure server-side handling—keys never exposed to the frontend

• Built-in error handling for missing or invalid keys

Multi-Agent Architecture:

• Each transformation type (paraphrasing, style, proficiency level, tone, readability) has its own dedicated AI agent with a specialized system prompt

• Agents work in sequence when users select multiple transformations

• Text chunking automatically handles large inputs (1200+ words) so no single agent gets overwhelmed

Modular Prompt System:

• All agent prompts stored in a configuration file you can easily edit

• Separate prompts for each proficiency level (B1, B2, C1, C2, native)

• Separate prompts for each tone (confident, cautious, friendly, professional)

• Separate prompts for each style (formal, casual, academic, conversational)

• Separate prompts for each readability level

• You can manually refine and update any prompt after deployment to fine-tune outputs

Production Features:

• Loading indicators so users know the app is processing

• Queue management for smooth handling of multiple requests

• Proper error feedback if something goes wrong

Technology stack: React for the frontend and Node.js for the backend.

Next Step After Purchasing This App — Add Your API Key:

1. Go to Open AI and create a free API key

2. Open your apps/api/.env file and replace the placeholder:

• OPENAI_API_KEY=sk-placeholder-key

3. Save and restart your app

• apps/api/.env now holds OPENAI_API_KEY=sk-placeholder-key + OPENAI_MODEL; the backend auto-switches to OpenAI once a real sk- key replaces the placeholder, with explicit 401/403 key errors. Keys stay server-side only.

Once you've added your real API key, the app will start transforming text instead of echoing it back. Try uploading some sample text, select a CEFR level and persona, and you'll see the agents working together to rewrite it.