AstroAgent Pipeline 🚀

Autonomous AI research system that takes ideas from conception to completion

What is this?

AstroAgent Pipeline is a goal-driven continuous research system that uses specialized AI agents to conduct astrophysics research automatically. Unlike traditional tools that require manual intervention at each step, this system runs autonomously until research goals are achieved.

Key features:

• 🎯 Goal-driven execution - auto-pauses when research targets are completed
• 🔄 Continuous pipeline - ideas flow through all stages without manual intervention
• 🤖 Smart AI agents - specialized for each research step
• 📚 Literature integration - real-time novelty checking against current papers
• 🌐 Real-time web dashboard - monitor progress and control pipeline
• 📊 Complete research audit trail - track every decision and result

🚀 Quick Start (2 minutes)

# 1. Clone and setup
git clone https://github.com/yourusername/astropipeline.git
cd astropipeline
pip install -r requirements.txt

# 2. Get free API keys and add to .env file
# ADS token: https://ui.adsabs.harvard.edu/user/settings/token
# OpenAI key: https://platform.openai.com/api-keys
echo "ADS_API_TOKEN=your_ads_token_here" > .env
echo "OPENAI_API_KEY=your_openai_key_here" >> .env

# 3. Start the autonomous research system!
python start.py --domains "exoplanets,atmospheres"

That's it!

• The system starts generating hypotheses, reviewing them, designing experiments, and executing research
• A web dashboard opens at http://localhost:8000 to monitor real-time progress
• The pipeline automatically pauses when the first idea completes the full research workflow
• Click any idea in the dashboard for detailed information
• Use the Pause/Resume button to control execution

🎯 Goal-driven approach: Instead of generating many disconnected ideas, the system focuses on advancing good ideas through the complete research pipeline until achievement criteria are met.

How it works

The system runs a continuous multi-agent research pipeline:

🔄 Continuous Pipeline Flow

1. Smart Hypothesis Generation - Creates ideas only when pipeline needs more work (avoids hypothesis overload)
2. Automatic Review - AI evaluates ideas for novelty, feasibility, and impact using current literature
3. Experiment Design - Approved ideas get detailed research protocols automatically
4. Execution Simulation - Research plans are validated and marked ready for implementation
5. Completion Tracking - Finished work is archived and counted toward goals

🎯 Goal-Driven Execution

• Default behavior: Auto-pause after 1 completed idea (configurable)
• Smart resource management: Only generates new hypotheses when pipeline needs more active work
• Focus on quality: Advances promising ideas through all stages before creating new ones
• Achievement-based stopping: Stops when research goals are met, not arbitrary time limits

🌐 Real-Time Dashboard

• Monitor all ideas progressing through pipeline stages
• See agent activity with live status indicators
• Pause/Resume/Stop controls for pipeline management
• Sortable, filterable tables with detailed idea information
• Click any idea for comprehensive details modal

What you can do right now

• ✅ Autonomous research execution - Full pipeline from idea to completion without manual intervention
• ✅ Real-time monitoring - Web dashboard shows live progress and agent activity
• ✅ Goal-driven operation - Automatically pauses when research targets achieved
• ✅ Smart idea management - Only generates new hypotheses when pipeline needs more work
• ✅ Interactive control - Pause/Resume/Stop pipeline via web interface
• ✅ Detailed idea exploration - Click any idea for comprehensive information
• 🚧 Actual experiment execution - Currently simulated, real data analysis coming soon

Usage Examples

# Default: Run continuous pipeline + web dashboard (auto-pause after 1 completed idea)
python start.py

# Run until 3 ideas are completed, then auto-pause
python start.py --complete-ideas 3

# Focus on specific research areas
python start.py --domains "stellar evolution,supernovae"

# Web dashboard only (useful if pipeline already running elsewhere)
python start.py web

# Pipeline only (no web interface)  
python start.py pipeline

# Traditional discrete mode (generate ideas once, then stop)
python start.py pipeline --mode discrete --count 5

Web Dashboard Features

• Real-time progress tracking - Watch ideas advance through pipeline stages
• Agent status monitoring - See which agents are active with visual indicators
• Pipeline controls - Pause, Resume, or Stop execution
• Interactive tables - Sort and filter ideas by status, score, domain
• Detailed idea modals - Click any row for full hypothesis, rationale, and research details

Requirements

• Python 3.9+
• Free ADS API token (for literature search) - Get yours here
• OpenAI API key (for AI agents) - Get yours here

Advanced Configuration

# Run with custom completion targets and time limits
python start.py --complete-ideas 5 --max-time 60  # 5 ideas OR 60 minutes max

# Adjust pipeline sensitivity (how many ideas to keep active)
python start.py --min-pipeline-size 2  # Keep 2-3 active ideas flowing

# Different research domains
python start.py --domains "galactic dynamics,dark matter,stellar formation"

# Use discrete mode for traditional batch processing
python start.py pipeline --mode discrete --count 10

Contributing

This is an active research project pushing the boundaries of autonomous AI research systems! We welcome:

• 🐛 Bug reports and feature requests
• 🤖 New agent implementations
• 📊 Integration with real astronomical data sources
• 🔬 Novel research pipeline strategies
• 📚 Documentation improvements

License

MIT License - feel free to use for research or commercial projects.

---

Autonomous AI research systems for the next generation of scientific discovery 🌟