swarmtrace 0.2.0

pytest for AI agents — trace, debug and catch regressions in LLM swarms

SwarmTrace 🐝

Pytest framework for AI Multi-Agents — trace, debug, and catch regressions in LLM swarms.

Built for the AMD Hackathon, SwarmTrace brings software engineering rigor to agentic AI systems. Test your multi-agent workflows like you test your code.

---

🎯 Why SwarmTrace?

LLM agents are unpredictable. SwarmTrace gives you:

• ✅ Trace agent reasoning — see every decision, tool call, and output
• ✅ Catch regressions — when agent behavior changes unexpectedly
• ✅ Debug failures — pinpoint exactly where the swarm broke
• ✅ CI/CD ready — integrates with any pytest pipeline

---

📦 Installation

git clone https://github.com/ravi3594444/swarmtrace.git
cd swarmtrace
pip install -r requirements.txt

🚀 Quick Start

from swarmtrace import SwarmTest, Agent

# Define your agent
agent = Agent(name="researcher", model="gpt-4")

# Write a test
def test_research_accuracy():
    result = agent.run("What is the capital of France?")
    assert "Paris" in result.output
    assert result.latency < 5.0  # seconds

---

🛠️ Tech Stack

• Python 3.10+
• pytest — test framework
• LangChain / custom agent orchestration
• OpenAI / Anthropic / Local LLM APIs

---

📝 License

MIT — built by Ravi Kumar