🧠 AI Prompt Analyzer & Tool Recommender

A Local-First AI Prompt Understanding Platform

Developer: Sabithulla

---

📌 Overview

AI Prompt Analyzer is an intelligent system that understands vague, unclear, or poorly written prompts and converts them into clear, actionable instructions.
It then analyzes user intent, calculates confidence, and recommends the most relevant AI tools for the task.

The project is designed to work locally with Ollama, ensuring:

• 💯 No API keys
• 🔒 Full privacy
• 💸 Zero cost

It also includes an Ask AI chatbot for user guidance and support (local mode).

---

✨ Why AI Prompt Analyzer?

Most AI tools expect users to write perfect prompts.

Real users don’t.

AI Prompt Analyzer bridges that gap by:

• ✏️ Rewriting unclear prompts
• 🧠 Detecting intent and domain
• 🛠️ Recommending the right AI tools
• 📊 Showing confidence with reasoning
• 💬 Assisting users through a chatbot

All features work offline using Ollama.

---

🖼️ Screenshots

🔍 Prompt Analysis & Tool Recommendations

🧰 Tool Detail Modal

💬 Ask AI Chatbot (Local)

---

🧪 Example Prompts

• Convert my voice recording into text
• Create a professional logo for my startup
• Write Python code for file upload
• Make Instagram reels automatically
• Summarize a research paper

---

✨ Key Features

• ✏️ Automatic prompt rewriting
• 🧠 Intent & domain extraction
• 🔍 Semantic tool matching (embeddings)
• 📊 Confidence scoring with breakdown
• 🛠️ AI tool recommendation engine
• 💬 Ask AI chatbot (local Ollama)
• 🎙️ Voice input (browser-based)
• ⛔ Stop analysis while running
• 🔒 Fully local & privacy-friendly

---

📊 Confidence Breakdown

Confidence is calculated using:

• Semantic similarity – 60%
• Intent match – 25%
• Domain match – 15%

This helps users understand why tools are recommended.

---

💬 Ask AI Chatbot (Local Only)

The chatbot helps users:

• Clarify prompts
• Ask follow-up questions
• Understand tool suggestions

⚠️ Note:
Chat works only in local mode because Ollama runs on your system.
Hosted versions show a fallback message.

---

🏗️ Architecture

Frontend (HTML + CSS + JavaScript)
↓
FastAPI Backend
↓
Prompt Analysis Pipeline (NumPy)
↓
Ollama (Local LLMs & Embeddings)

---

🛠️ Tech Stack

Backend

• Python 3.9+
• FastAPI
• Uvicorn
• NumPy
• Requests
• Ollama

Frontend

• HTML
• CSS
• JavaScript
• Web Speech API

AI Models (Ollama)

• qwen2:0.5b → Chat
• nomic-embed-text → Embeddings

✅ Free
✅ Offline
✅ No API keys

---

📂 Project Structure

ai-tools-platform/ ├── frontend/ │
└── index.html │ ├── src/ │
├── api.py │
├── chat_api.py │
├── final_pipeline_numpy.py │ ├── intent_extractor.py │ ├── prompt_rewriter.py │ ├── filter_tools.py │
└── moderation.py │ ├── data/ │ ├── tools_seed.json │ ├── tool_embeddings.npy │ ├── tool_ids.json │
└── bad_prompts.log │ ├── screenshots/ │ ├── prompt-analysis.png │ ├── tool-details.png │
├── ask-ai-chat.png │
└── voice-input.png │ └── README.md

---

⚙️ Installation & Setup (Local)

Requirements

• Python 3.9+
• Ollama installed locally

Install Ollama

• https://ollama.com/download

Pull Required Models

ollama pull qwen2:0.5b
ollama pull nomic-embed-text
ollama serve

---

### How to Run
### Backend

pip install -r requirements.txt
uvicorn src.api:app --reload

Backend runs at:
http://127.0.0.1:8000

### Frontend
cd frontend
python -m http.server 5500

Open in browser:
http://127.0.0.1:5500

---

### Developer

Valtry
Student • AI Builder • Problem Solver

Built with persistence, debugging, and real-world prompt understanding.

---

### Support

If you find this useful:
⭐ Star the repository
🍴 Fork it
🧠 Extend it

---

📜 License
This project is released under the MIT License.
You are free to use, modify, and distribute it.

---