0.1.0

---

🎯 The Problem

Multi-agent systems are opaque black boxes. When your 5-agent pipeline produces a bad result, you have no way to see:

• Which agent said what to whom
• Why things went wrong
• Where the breakdown happened
• How much it cost

Debugging is impossible. Optimization is guesswork.

✨ The Solution

pip install agentwire
agentwire start

# Wrap your agents (one line each)
researcher = aw.wrap(ResearchAgent(), name="researcher")
writer = aw.wrap(WriterAgent(), name="writer")

# Run your pipeline
with aw.session("blog-post-run"):
    facts = researcher.run("Find quantum computing breakthroughs")
    post = writer.run(facts)

Open http://localhost:7433 → See everything in real-time.

---

🚀 Features

---

⚡ Quick Start

Installation

pip install agentwire

Start the Server

agentwire start

This starts:

• 🌐 REST API at http://localhost:7433/api
• 🔌 WebSocket at ws://localhost:7433/ws
• 📊 Dashboard at http://localhost:7433

Wrap Your Agents

import agentwire as aw

# Configure once
aw.configure(bus_url="http://localhost:7433")

# Wrap any agent (works with any class)
researcher = aw.wrap(ResearchAgent(), name="researcher")
writer = aw.wrap(WriterAgent(), name="writer")
reviewer = aw.wrap(ReviewerAgent(), name="reviewer")

# Run with session grouping
with aw.session("blog-post-run-42", name="Blog Post Generation"):
    facts = researcher.run("Find quantum computing breakthroughs")
    draft = writer.run(facts)
    final = reviewer.run(draft)

View in Dashboard

Open http://localhost:7433 to see:

• 📝 Feed View - Real-time message stream with auto-scroll
• 🕸️ Graph View - Agent topology with interactive nodes
• 📊 Stats - Messages, tokens, cost tracking
• ⏮️ Replay - Step through session with timeline controls

---

🎨 What Makes It Different

AgentWire focuses on the message bus abstraction - what agents say to each other - not just LLM calls.

---

📚 Examples

LangChain Research Pipeline

from agentwire.integrations.langchain import AgentWireCallback
import agentwire as aw

aw.configure(bus_url="http://localhost:7433")

with aw.session("research-run"):
    planner = aw.wrap(PlannerAgent(), name="planner")
    researcher = aw.wrap(ResearchAgent(), name="researcher")
    summarizer = aw.wrap(SummarizerAgent(), name="summarizer")
    
    plan = planner.run("Quantum computing breakthroughs 2024")
    findings = researcher.run(plan)
    summary = summarizer.run(findings)

→ See full example

AutoGen Coding Team

from agentwire.integrations.autogen import wire_autogen_agent
import agentwire as aw

with aw.session("coding-session"):
    orchestrator = aw.wrap(OrchestratorAgent(), name="orchestrator")
    coder = aw.wrap(CoderAgent(), name="coder")
    reviewer = aw.wrap(ReviewerAgent(), name="reviewer")
    
    assignment = orchestrator.run("Write Fibonacci function")
    code = coder.run(assignment)
    review = reviewer.run(code)

→ See full example

Raw API Pipeline

import agentwire as aw

with aw.session("data-pipeline"):
    fetcher = aw.wrap(DataFetcher(), name="fetcher")
    processor = aw.wrap(DataProcessor(), name="processor")
    analyzer = aw.wrap(DataAnalyzer(), name="analyzer")
    reporter = aw.wrap(Reporter(), name="reporter")
    
    data = fetcher.run("production-db")
    processed = processor.run(data)
    analysis = analyzer.run(processed)
    report = reporter.run(analysis)

→ See full example

---

🔧 Integrations

LangChain

from agentwire.integrations.langchain import AgentWireCallback

agent = initialize_agent(
    tools=[...],
    llm=llm,
    callbacks=[AgentWireCallback(agent_name="researcher")]
)

AutoGen

from agentwire.integrations.autogen import wire_autogen_agent

wire_autogen_agent(my_agent, name="assistant", session_id="run-1")

CrewAI

from agentwire.integrations.crewai import wire_crew

my_crew = Crew(agents=[...], tasks=[...])
wire_crew(my_crew, session_id="crew-run")
result = my_crew.kickoff()

---

🎮 CLI Commands

# Start server
agentwire start                    # Default: port 7433
agentwire start --port 8000        # Custom port
agentwire start --no-dashboard     # API-only mode

# Check status
agentwire status                   # Show server status and stats

# Clear data
agentwire clear                    # Clear all (with confirmation)
agentwire clear --session abc123   # Clear specific session
agentwire clear --force            # Skip confirmation

# Stop server
agentwire stop

# Docker
agentwire docker up                # Start with Docker Compose
agentwire docker down              # Stop containers

---

📖 Documentation

• Quick Start Guide - Get started in 5 minutes
• API Reference - Complete REST & WebSocket API
• SDK Documentation - Python SDK reference
• Integration Guides - Framework integrations
• Dashboard Guide - Using the web interface
• Examples - Working code examples

---

🏗️ Architecture

┌─────────────────┐
│   Agent Code    │  Your Python code
│   (SDK)         │  aw.wrap(), aw.session()
└────────┬────────┘
         │ emit()
         ↓
┌─────────────────┐
│   REST API      │  FastAPI server
│   POST /api/    │  Store + broadcast
└────────┬────────┘
         │
    ┌────┴────┐
    ↓         ↓
┌────────┐ ┌──────────┐
│ SQLite │ │WebSocket │  Real-time
│  DB    │ │Broadcast │  to clients
└────────┘ └─────┬────┘
                 ↓
         ┌───────────────┐
         │ React         │  Dashboard
         │ Dashboard     │  Feed + Graph
         └───────────────┘

---

🧪 Testing

# Run all tests (54 passing)
pytest tests/ -v

# Run specific test suite
pytest tests/test_bus.py -v
pytest tests/test_integrations.py -v

# Run with coverage
pytest tests/ --cov=agentwire --cov-report=html

---

🛠️ Development

Setup

# Clone repository
git clone https://github.com/DanixMP/AgentWire.git
cd agentwire

# Install in development mode
pip install -e .
pip install -r requirements-dev.txt

# Run tests
pytest tests/

# Start dashboard dev server
cd dashboard
npm install
npm run dev

Project Structure

agentwire/
├── agentwire/              # Python package
│   ├── models.py           # Data models
│   ├── store.py            # Database layer
│   ├── bus.py              # FastAPI server
│   ├── emitter.py          # Message emission
│   ├── wrapper.py          # Agent wrapping
│   ├── session.py          # Session management
│   ├── cli.py              # CLI tool
│   └── integrations/       # Framework integrations
├── dashboard/              # React dashboard
│   └── src/
│       ├── components/     # React components
│       ├── hooks/          # Custom hooks
│       └── types.ts        # TypeScript types
├── examples/               # Example scripts
├── tests/                  # Test suite
└── docs/                   # Documentation

---

🎯 Roadmap

✅ Completed

• Core message bus and storage
• Real-time WebSocket broadcasting
• React dashboard with feed and graph
• Session replay with timeline
• CLI tool
• Framework integrations (LangChain, AutoGen, CrewAI)
• Examples and documentation

🚧 Planned

• Postgres support for production
• Multi-session comparison
• Export sessions as JSON/CSV
• Custom message types
• Webhook notifications
• Team collaboration features
• Cloud deployment guides

---

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

---

📄 License

MIT License - see LICENSE file for details.

---

🙏 Acknowledgments

• Built with FastAPI, React, and D3.js
• Inspired by the need for better multi-agent observability
• Thanks to the LangChain, AutoGen, and CrewAI communities

---

📞 Support

• 📖 Documentation
• 💬 GitHub Issues
• 🐦 Twitter
• 📧 Email

---