🌌 Astera

Autonomous Deep Research Workspace & Evidence Synthesis Engine

Research anything. Astera searches the web, reads trusted sources, reconciles evidence, and writes comprehensive research reports.

---

📌 Overview

Astera is an open-source deep research application designed to turn complex research prompts into clear, highly structured, evidence-backed reports.

Unlike standard conversational search bots that summarize snippet search results, Astera behaves as an autonomous research team:

1. Plans Research Strategy: Breaks a prompt into multiple distinct search angles.
2. Discovers & Scores Sources: Ranks web domains by authority and trust.
3. Scrapes Full Web Pages: Reads full text rather than relying on brief snippets.
4. Reconciles Evidence: Compares facts, removes duplicate citations, and drafts report sections.
5. Evaluates Completeness: Runs a Critic agent pass to verify quality before delivering the final report.

Astera supports both anonymous users (zero storage, transient sessions) and authenticated users (persistent cloud history, multi-turn follow-up conversations, search, and title auto-generation).

---

⚡ Why Astera?

Traditional search engines leave you clicking through dozens of tabs, while standard LLMs are prone to hallucinating facts or citing superficial snippets.

How Astera is different:

• Evidence-First: Every report section is built directly from scraped web content.
• Multi-Turn Workspace: Continue any research session by asking follow-up questions beneath the original report, exactly like ChatGPT.
• Authority Ranking: Filters out clickbait and content farms by scoring domain authority (.gov, .edu, trusted journalism, academic portals).
• Cloud-Synced History: Authenticated users can restore, search, pin, rename, or archive past research sessions effortlessly.

---

✨ Features

• 🔍 Multi-Query Expansion: Automatically plans 3-4 distinct search queries per topic.
• 🌐 Deep Web Reading: Scrapes full article content from top authority sources.
• ⚡ Real-Time Progress Streaming: Emits live Server-Sent Events (SSE) detailing every step (Searching, Reading X/Y, Writing, Checking).
• 💬 Continuous Multi-Turn Chat: Follow-up prompts append seamlessly without losing past research reports.
• 🔐 Production Auth.js (NextAuth v5): OAuth sign-in via Google & GitHub, with zero-friction anonymous access.
• 🗄️ Supabase PostgreSQL & Prisma ORM: Scalable schema built for conversations, messages, sessions, sources, bookmarks, and tags.
• 📱 Mobile-First & Accessible: Fully responsive drawer sidebar, touch controls, keyboard navigation, and ARIA live regions.
• 🎨 Modern Aesthetics: Sleek dark mode, glassmorphism, auto-expanding textarea, and domain favicons.

---

🏗️ Architecture

Astera connects a high-performance FastAPI multi-agent backend with a Next.js App Router frontend.

graph TD
    User["👤 User / Browser"] -->|Next.js App Router| Frontend["💻 Next.js Frontend (React 19)"]
    Frontend -->|Auth.js v5| Auth["🔐 Auth.js (Google / GitHub)"]
    Auth -->|Prisma Client| DB[("🗄️ Supabase PostgreSQL")]
    
    Frontend -->|SSE Progress / GET| Backend["⚡ FastAPI Backend (Python)"]
    Backend -->|Multi-Query Planner| Planner["🎯 Query Generator"]
    Planner -->|Search API| Tavily["🔍 Tavily Web Search"]
    Tavily -->|Domain Authority Ranking| Scraper["🌐 Playwright / BeautifulSoup Scraper"]
    
    Scraper -->|Full Text Context| Writer["✍️ Writer Agent (LLM)"]
    Writer -->|Draft Report| Critic["🧐 Critic Review Agent"]
    Critic -->|Final Verified Result| Frontend

Loading

---

📁 Repository Structure

Astera/
├── .github/                      # Issue & PR templates
│   ├── ISSUE_TEMPLATE/
│   │   ├── bug_report.md
│   │   └── feature_request.md
│   └── PULL_REQUEST_TEMPLATE.md
├── agents/                       # Multi-agent definitions (Planner, Scraper, Writer, Critic)
├── chains/                       # LangChain execution chains
├── models/                       # Multi-provider LLM connector (Groq, Cerebras, Gemini, Mistral)
├── pipelines/                    # Deep research orchestrator & SSE generator
├── tools/                        # Web search & deep scraping tools
├── api.py                        # FastAPI production application
├── ui/                           # Next.js App Router frontend
│   ├── app/                      # App router pages, routes & metadata
│   ├── components/               # React UI components (Sidebar, SearchInput, ChatContainer)
│   ├── hooks/                    # Custom React hooks (useConversation, useResearch)
│   ├── lib/                      # Auth.js, Prisma, API clients, TypeScript definitions
│   └── prisma/
│       └── schema.prisma         # Production Supabase PostgreSQL schema
├── CONTRIBUTING.md               # Contribution guidelines
├── LICENSE                       # MIT License
├── SECURITY.md                   # Security reporting policy
└── requirements.txt              # Backend Python dependencies

---

🛠️ Tech Stack

Layer	Technologies
Frontend Framework	Next.js 16 (App Router), React 19, TypeScript, Tailwind CSS, Framer Motion
Authentication	Auth.js (NextAuth v5), Google OAuth, GitHub OAuth
Database & ORM	Supabase PostgreSQL, Prisma ORM (v6)
Backend API	Python 3.11+, FastAPI, Uvicorn, SSE Streaming
Agent Orchestration	LangChain, Pydantic, Custom Evidence Pipeline
Web Search & Scraping	Tavily API, BeautifulSoup4, Playwright
Deployment	Vercel (Frontend), Railway (Backend), Supabase (Database)

---

🚀 Local Installation & Setup

Prerequisites

• Node.js: v20+ and npm
• Python: v3.11+
• Database: PostgreSQL connection URI (e.g. Supabase)

1. Clone the Repository

git clone https://github.com/your-username/Astera.git
cd Astera

2. Backend Setup

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Configure environment variables
cp .env.example .env

Fill your API keys in .env:

TAVILY_API_KEY=tvly-your-key
GROQ_API_KEY=gsk_your_key
GEMINI_API_KEY=your_key
ALLOWED_ORIGINS=http://localhost:3000

Start the backend server:

python api.py

Backend runs on http://localhost:8000 (Docs available at http://localhost:8000/docs).

3. Frontend Setup

Open a new terminal window:

cd ui

# Install dependencies
npm install

# Setup local environment
cp .env.example .env.local

Configure ui/.env.local:

NEXT_PUBLIC_API_URL=http://localhost:8000
DATABASE_URL=postgresql://postgres:password@localhost:5432/astera
DIRECT_URL=postgresql://postgres:password@localhost:5432/astera

AUTH_SECRET=your-random-auth-secret
AUTH_GOOGLE_ID=your-google-client-id
AUTH_GOOGLE_SECRET=your-google-client-secret
AUTH_GITHUB_ID=your-github-client-id
AUTH_GITHUB_SECRET=your-github-client-secret

Generate Prisma Client & run migrations:

npm exec prisma generate

Start Next.js dev server:

npm run dev

Frontend runs on http://localhost:3000.

---

🌐 Production Deployment

Frontend (Vercel)

1. Push your repository to GitHub.
2. Import project into Vercel and select root directory ui.
3. Set Environment Variables (NEXT_PUBLIC_API_URL, DATABASE_URL, AUTH_SECRET, AUTH_GOOGLE_ID, AUTH_GOOGLE_SECRET, etc.).
4. Deploy!

Backend (Railway)

1. Deploy root directory to Railway using the included Dockerfile or Procfile.
2. Add environment variables (TAVILY_API_KEY, GROQ_API_KEY, GEMINI_API_KEY, ALLOWED_ORIGINS).

---

🗺️ Roadmap

• Multi-query research strategy & authority scoring
• Real-time SSE progress streaming
• Auth.js (NextAuth v5) Google & GitHub OAuth integration
• Supabase PostgreSQL & Prisma ORM persistent storage
• Continuous multi-turn conversation feed & auto-expanding input
• Bookmark specific report sections
• Group conversations into Folders & Collections
• One-click PDF & Markdown export
• Public shared conversation links

---

📄 License

Distributed under the MIT License. See LICENSE for details.

---

🙏 Acknowledgements

• Tavily AI for precision search API.
• Vercel & Next.js for modern web primitives.
• Auth.js for open-source authentication.
• Prisma & Supabase for database infrastructure.