Firesearch - AI-Powered Deep Research Tool

Version 1.1 - Comprehensive web research powered by Firecrawl and LangGraph

🚀 What's New in v1.1

💻 Advanced Settings Page - Comprehensive configuration with environment detection
🔧 Multiple LLM Provider Support - OpenAI, Anthropic, OpenRouter, Google Gemini, Groq, Grok AI, and Deepseek
🎨 Enhanced UI/UX - Reorganized navigation, floating contribute button, improved visuals
🗺️ Comprehensive Roadmap - Chat Page 2.0, advanced features, and planned expansions
🔒 Better Security - Improved API key handling with environment variable precedence
📱 Responsive Design - Optimized for all screen sizes

🛠️ Technologies

Firecrawl: Multi-source web content extraction and search
LangGraph: Intelligent agent workflow orchestration
Multiple LLM Providers: OpenAI GPT-4o, Anthropic Claude, Google Gemini, and more
Next.js 15: Modern React framework with App Router and Turbopack
TypeScript: Type-safe development with comprehensive error handling

🔑 API Key Configuration

Required Keys

Service	Purpose	Get Key
Firecrawl	Web scraping and content extraction	firecrawl.dev/app/api-keys
LLM Provider	AI reasoning and synthesis	See options below

Supported LLM Providers

Provider	Models Available	Get API Key
OpenAI	GPT-4o, GPT-4o-mini, GPT-3.5-turbo	platform.openai.com/api-keys
Anthropic	Claude 3.5 Sonnet, Claude 3 Haiku	console.anthropic.com
Google Gemini	Gemini Pro, Gemini Flash	ai.google.dev
OpenRouter	Access to 100+ models	openrouter.ai/keys
Groq	Ultra-fast Llama, Mixtral, Gemma	console.groq.com
Grok AI	xAI's Grok models	x.ai
Deepseek	Advanced reasoning models	platform.deepseek.com

⚡ Quick Start

Using pnpm (Recommended)

# Clone the repository
git clone https://github.com/mendableai/firesearch.git
cd firesearch

# Install dependencies
pnpm install

# Create environment file
cp .env.example .env.local

# Add your API keys to .env.local
FIRECRAWL_API_KEY=your_firecrawl_key
OPENAI_API_KEY=your_openai_key

# Start development server
pnpm dev

Alternative Package Managers

# Using npm
npm install && npm run dev

# Using yarn
yarn install && yarn dev

Using Docker

docker build -t firesearch .
docker run -p 3000:3000 --env-file .env.local firesearch

🎯 Key Features

Core Capabilities

🧠 Smart Query Decomposition - Breaks complex questions into focused sub-queries
🔍 Multi-Source Search - Searches multiple sources simultaneously via Firecrawl
✅ Answer Validation - Verifies source relevance (0.7+ confidence threshold)
🔄 Intelligent Retry Logic - Alternative search strategies for incomplete answers
📄 Real-time Progress - Live updates as searches complete
📚 Full Citations - Every fact linked to its original source
💬 Context Memory - Follow-up questions maintain conversation history

User Experience

⚙️ Advanced Settings - Comprehensive configuration with environment detection
🎨 Modern UI - Clean, responsive design with intuitive navigation
🔒 Privacy-First - API keys stored securely in local storage
🌐 Cross-Platform - Works on desktop, tablet, and mobile devices
🚀 Fast Performance - Optimized with Next.js 15 and Turbopack

🗺️ Future Roadmap

Chat Page 2.0

💬 Discussion the Research - Interactive discussions about search results
👁️ View Active Model - Real-time display of models being used
📖 Search History - Browse and revisit previous searches
📊 Search Analytics - Insights into search patterns and cost tracking
💰 Real-Time Cost Calculator - Live cost tracking for each search request
✅ Manual Settings Confirmation - Explicit save actions with confirmation dialogs

Interface & Theme Enhancements

🌙 Dark Mode & Light Mode - Enhanced theme system with user preferences
📱 Responsive Design - Optimized mobile and tablet experience
🎨 Accessibility - WCAG 2.1 AA compliance and keyboard navigation

Advanced Features

🌍 Multi-Language Support - Search and interface localization
📤 Export & Share - PDF export and sharing research findings
🎤 Voice Search - Speech-to-text input capabilities
🤖 Custom Search Agents - Specialized research agents for specific domains
🔌 Browser Extensions - Quick access from any webpage
📊 Data Visualization - Charts and graphs for research insights

LLM Provider Expansions

🏠 Local LLM Support - Ollama, LM Studio, and other local inference servers
🔒 Self-hosted Firecrawl - Connect to your own Firecrawl instance for enhanced security
🚀 Emerging Providers - Support for specialized reasoning engines and new AI platforms

🏗️ Architecture

How It Works

flowchart TB
    Query["User Query:<br/>'Compare latest iPhones'"]:::query
    
    Query --> Break
    
    Break["🔍 Query Decomposition"]:::primary
    
    subgraph SubQ["🌐 Sub-Questions"]
        S1["iPhone 15 Pro specs"]:::search
        S2["iPhone 15 pricing"]:::search
        S3["iPhone 15 vs 14 comparison"]:::search
    end
    
    Break --> SubQ
    
    subgraph FC["🔥 Firecrawl Searches"]
        FC1["Search Apple.com<br/>TechCrunch"]:::firecrawl
        FC2["Search GSMArena<br/>The Verge"]:::firecrawl
        FC3["Search CNET<br/>MacRumors"]:::firecrawl
    end
    
    S1 --> FC1
    S2 --> FC2
    S3 --> FC3
    
    subgraph Valid["✅ Validation"]
        V1["Specs found ✓ (0.95)"]:::good
        V2["Pricing found ✓ (0.85)"]:::good
        V3["Comparison ❌ (0.4)"]:::bad
    end
    
    FC --> Valid
    
    Valid --> Synthesis["🤖 LLM Synthesis"]:::synthesis
    
    Synthesis --> Answer["📄 Cited Response<br/>+ Follow-ups"]:::answer
    
    classDef query fill:#ff8c42,stroke:#ff6b1a,stroke-width:3px,color:#fff
    classDef primary fill:#3a4a5c,stroke:#2c3a47,stroke-width:2px,color:#fff
    classDef search fill:#4caf50,stroke:#388e3c,stroke-width:2px,color:#fff
    classDef firecrawl fill:#ff6b1a,stroke:#ff4500,stroke-width:2px,color:#fff
    classDef good fill:#4caf50,stroke:#388e3c,stroke-width:2px,color:#fff
    classDef bad fill:#f44336,stroke:#d32f2f,stroke-width:2px,color:#fff
    classDef synthesis fill:#9c27b0,stroke:#7b1fa2,stroke-width:2px,color:#fff
    classDef answer fill:#3a4a5c,stroke:#2c3a47,stroke-width:3px,color:#fff

Processing Pipeline

🔍 Query Analysis - Understanding user intent and complexity
📝 Decomposition - Breaking into focused sub-questions
🌐 Multi-Search - Parallel searches via Firecrawl API
📄 Content Extraction - Markdown content from relevant sources
✅ Validation - Confidence scoring for answer completeness
🔄 Retry Logic - Alternative strategies for gaps
🤖 Synthesis - LLM combines findings into comprehensive response
📚 Citation - Full source attribution and follow-up generation

⚙️ Configuration

Customize search behavior in the Settings page or via lib/config.ts:

export const SEARCH_CONFIG = {
  // Search Parameters
  MAX_SEARCH_QUERIES: 12,        // Maximum parallel searches
  MAX_SOURCES_PER_SEARCH: 4,     // Sources per search query
  MAX_SOURCES_TO_SCRAPE: 3,      // Additional content scraping
  
  // Quality Control
  MIN_CONTENT_LENGTH: 100,       // Minimum viable content length
  MIN_ANSWER_CONFIDENCE: 0.7,    // Answer completeness threshold
  
  // Performance
  MAX_RETRIES: 2,                // Retry attempts for failures
  MAX_SEARCH_ATTEMPTS: 2,        // Alternative search strategies
  SCRAPE_TIMEOUT: 15000,         // Content extraction timeout
  
  // Response Generation
  SUMMARY_CHAR_LIMIT: 100,       // Source summary length
  ENABLE_FOLLOW_UPS: true,       // Generate follow-up questions
} as const;

🧪 Example Queries

Technology Research

"Compare the latest iPhone, Samsung Galaxy, and Google Pixel flagship features"
"What are the key differences between ChatGPT-4 and Claude 3.5 Sonnet?"
"NVIDIA RTX 4090 vs RTX 4080 Super performance benchmarks"

Business Intelligence

"Who are the founders of Firecrawl and what's their background?"
"Latest funding rounds in AI infrastructure companies 2024"
"Tesla's Q4 2024 earnings vs analyst expectations"

Academic Research

"Recent breakthroughs in quantum computing error correction"
"Climate change impact on arctic ice coverage 2020-2024"
"Effectiveness of different learning management systems for remote education"

🤝 Contributing

We welcome contributions! Here's how to get started:

Fork the repository
Create a feature branch: git checkout -b feature/amazing-feature
Make your changes and ensure tests pass
Commit your changes: git commit -m "Add amazing feature"
Push to your branch: git push origin feature/amazing-feature
Open a Pull Request

Development Guidelines

Follow TypeScript best practices
Add tests for new features
Update documentation as needed
Use conventional commits for messages
Ensure accessibility compliance

📚 API Documentation

Firecrawl Integration

Firesearch leverages Firecrawl's /search endpoint for comprehensive web research:

// Search with content extraction
const response = await fetch('/api/search', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    query: "iPhone 16 specifications",
    limit: 8,
    scrapeOptions: {
      formats: ["markdown"]
    }
  })
});

LLM Provider Integration

Support for multiple providers with unified interface:

// Dynamic provider switching
const llmResponse = await queryLLM({
  provider: "openai", // or "anthropic", "gemini", etc.
  model: "gpt-4o",
  messages: [...],
  apiKey: getApiKey(provider)
});

🏠 Self-Hosting & Privacy

Firesearch is designed with privacy and control in mind. Future versions will support:

🔒 Self-hosted Firecrawl Integration

Connect to your own Firecrawl instance for:

Enhanced Security - All data remains within your infrastructure
Compliance - Meet strict regulatory requirements (GDPR, HIPAA, SOC2)
Cost Control - Avoid per-request API fees for high-volume usage
Customization - Tailor scraping behavior for specific needs

🏠 Local LLM Support

Run AI models locally for complete privacy:

Ollama Integration - Support for Llama, Mistral, and other open models
LM Studio Support - Easy local model management and inference
Custom Endpoints - Connect to any OpenAI-compatible local server
Zero Data Sharing - All processing happens on your hardware

🛡️ Benefits of Self-Hosting

Complete Privacy - No data ever leaves your environment
Cost Optimization - Reduce API costs for high-volume usage
Regulatory Compliance - Meet enterprise security requirements
Custom Configuration - Fine-tune behavior for specific use cases

🔧 Deployment

Vercel (Recommended)

Click the "Deploy with Vercel" button above
Add your environment variables in the Vercel dashboard
Deploy automatically on every push to main

Self-Hosting

# Build for production
pnpm build

# Start production server
pnpm start

# Or use Docker
docker build -t firesearch .
docker run -p 3000:3000 firesearch

Environment Variables

# Required
FIRECRAWL_API_KEY=your_firecrawl_key
OPENAI_API_KEY=your_openai_key

# Optional (can be set in Settings page)
ANTHROPIC_API_KEY=your_anthropic_key
GOOGLE_API_KEY=your_google_key
OPENROUTER_API_KEY=your_openrouter_key
GROQ_API_KEY=your_groq_key
GROK_API_KEY=your_grok_key
DEEPSEEK_API_KEY=your_deepseek_key

# Self-hosting Options (Coming Soon)
FIRECRAWL_API_URL=http://localhost:3002  # For self-hosted Firecrawl
OLLAMA_BASE_URL=http://localhost:11434   # For local Ollama models
LM_STUDIO_BASE_URL=http://localhost:1234 # For LM Studio local server

📄 License

MIT License - see LICENSE file for details.

🙏 Acknowledgments

Firecrawl for powerful web scraping capabilities
LangChain for LangGraph agent framework
Vercel for seamless deployment platform
Open source community for continuous inspiration and contributions

Built with ❤️ by the Firesearch Team
Powered by Firecrawl and LangGraph

Name		Name	Last commit message	Last commit date
Latest commit History 13 Commits
app		app
components/ui		components/ui
hooks		hooks
lib		lib
public		public
.gitignore		.gitignore
README.md		README.md
components.json		components.json
eslint.config.mjs		eslint.config.mjs
next.config.ts		next.config.ts
package.json		package.json
pnpm-lock.yaml		pnpm-lock.yaml
postcss.config.mjs		postcss.config.mjs
tailwind.config.ts		tailwind.config.ts
tsconfig.json		tsconfig.json
vercel.json		vercel.json

ImTheMars/firesearch

Folders and files

Latest commit

History

Repository files navigation