AstraMindChat 🌌

A multi-agent AI chat application with a React frontend and FastAPI backend, supporting multiple LLM providers, custom agents, analytics, and prompt optimization tools.

---

📋 Table of Contents

• Overview
• Features
• Tech Stack
• Project Structure
• Getting Started
  • Prerequisites
  • Backend Setup
  • Frontend Setup
• Configuration
• Built-in Agents
• API Reference
• Usage Guide

---

Overview

AstraMindChat is a full-stack AI chat platform that lets users interact with specialized AI agents powered by Google Gemini or Groq (Llama 3). It features a live analytics dashboard, exportable chat history, prompt templates, a Prompt Reviewer panel, and a custom agent builder — all wrapped in a clean dark-mode UI.

---

✨ Features

Feature	Description
🤖 Multi-Agent Chat	6 built-in AI agents, each with a specialized system prompt
🧠 Dual LLM Providers	Switch between Google Gemini and Groq (Llama 3)
🔍 Prompt Reviewer	Real-time AI-powered prompt quality analysis via Groq
✍️ Prompt Generator	Rewrites your rough input into an optimized prompt
📊 Analytics Dashboard	Tracks total prompts, agent usage, provider usage, and mode breakdowns
📁 Chat Export	Download conversation history as JSON or TXT
📝 Prompt Templates	Pre-built templates to speed up common queries
🛠️ Custom Agent Builder	Create, save, and delete your own agents with custom system prompts
🌙 Dark / Light Mode	Persistent theme toggle
💾 Persistent Storage	Chat history and analytics saved to a SQLite database via FastAPI

---

🛠️ Tech Stack

Frontend

• React 19 with Vite
• Lucide React for icons
• Vanilla CSS (custom design system with CSS variables)
• localStorage for client-side persistence (theme, API keys, analytics)

Backend

• FastAPI — REST API server
• SQLModel — ORM on top of SQLAlchemy
• SQLite — Embedded database (astramind.db)
• Pydantic — Request/response validation

AI / LLM

• Google Gemini API (default provider)
• Groq API — Llama 3, used for the Prompt Reviewer panel

---

📁 Project Structure

internship-eval-proj/
├── backend/
│   └── venv/
│       └── main.py          # FastAPI app — all routes and DB models
│
└── frontend/
    ├── index.html
    ├── vite.config.js
    ├── package.json
    └── src/
        ├── App.jsx                  # Root component, global state
        ├── main.jsx
        ├── agents/
        │   └── agents.jsx           # Built-in agent definitions + merge util
        ├── components/
        │   ├── analytics/
        │   │   └── AnalyticsView.jsx
        │   ├── chat/
        │   │   ├── ChatWindow.jsx
        │   │   ├── InputBox.jsx
        │   │   ├── MessageBubble.jsx
        │   │   ├── PromptGenerator.jsx
        │   │   └── PromptReviewerPanel.jsx
        │   ├── header/
        │   │   └── Header.jsx
        │   ├── settings/
        │   │   └── Settings.jsx
        │   └── sidebar/
        │       └── Sidebar.jsx
        ├── hooks/
        │   └── useLocalStorage.js
        ├── styles/
        │   └── global.css
        ├── templates/
        │   └── templates.js
        └── utils/
            └── exportUtils.js

---

🚀 Getting Started

Prerequisites

• Node.js v18+ and npm
• Python 3.10+
• A Google Gemini API Key → Get one here
• (Optional) A Groq API Key for the Prompt Reviewer → Get one here

---

Backend Setup

# 1. Navigate to the backend venv directory
cd backend/venv

# 2. Install dependencies
pip install fastapi uvicorn sqlmodel

# 3. Start the server
uvicorn main:app --reload --port 8000

The backend will be available at http://localhost:8000.
The SQLite database (astramind.db) is created automatically on first run.

---

Frontend Setup

# 1. Navigate to the frontend directory
cd frontend

# 2. Install dependencies
npm install

# 3. Start the development server
npm run dev

The app will be available at http://localhost:5173.

---

⚙️ Configuration

All configuration is done from the Settings view inside the app:

Setting	Options	Description
Tone	Friendly, Formal, Casual, Creative, Technical	Controls the AI's conversational tone
Response Style	Short, Medium, Long, Detailed	Controls response length
Model Mode	Online, Offline	Switches between live API and offline mode
LLM Provider	Gemini (Default), Groq (Llama 3)	Selects which AI provider to use
Gemini API Key	(your key)	Stored in localStorage
Groq API Key	(your key)	Used for the Prompt Reviewer panel

---

🤖 Built-in Agents

Agent	ID	Description
Astra Chat	general	Friendly general-purpose assistant
DevCore	code	Elite coding assistant for clean, optimized code
Scribe	writer	Expert writing and rewriting assistant
LogicNode	math	Step-by-step mathematical reasoning
Strategist	planner	Structured planning and goal breakdown
PromptGenerator	prompt	Rewrites user queries into optimized prompts

You can also create custom agents from the Settings panel by providing a name, system prompt, color, and emoji icon. Custom agents are persisted in the backend database.

---

📡 API Reference

The backend runs on http://localhost:8000.

Messages

Method	Endpoint	Description
POST	/api/message	Save a chat message
GET	/api/history	Retrieve all chat messages

Analytics

Method	Endpoint	Description
POST	/api/analytics	Save analytics data
GET	/api/analytics	Retrieve analytics data

Custom Agents

Method	Endpoint	Description
POST	/api/agents	Create a new custom agent
GET	/api/agents	List all custom agents
DELETE	/api/agents/{id}	Delete a custom agent by ID

System

Method	Endpoint	Description
GET	/	Health check
DELETE	/api/reset	Clear all messages, analytics, and custom agents

---

📖 Usage Guide

1. Enter your API key in Settings → Gemini API Key.
2. Pick an agent from the dropdown in the chat input (or leave on "Auto").
3. Type your message and press Send (or Enter).
4. Use Generate Prompt to let AI refine your input before sending.
5. The Prompt Reviewer Panel (requires Groq API Key) gives real-time feedback on your draft.
6. Browse the Sidebar to load prompt templates or export/reset your chat.
7. Switch to the Analytics view to see your usage breakdown.