agentic-fleet 0.6.6

Multi-agent orchestration system built with Microsoft Agent Framework's Magentic Fleet pattern

AgenticFleet

A self-optimizing multi-agent orchestration system combining DSPy for intelligent task routing with Microsoft agent-framework for robust execution.

Note: APIs may change between minor versions. Pin a version for production.

Quick Start

# Install
git clone https://github.com/Qredence/agentic-fleet.git && cd agentic-fleet
uv sync  # or: pip install agentic-fleet

# Configure
cp .env.example .env
# Edit .env: set OPENAI_API_KEY (required), TAVILY_API_KEY (optional for web search)

# Run
agentic-fleet run -m "Research the latest AI advances" --verbose

What It Does

AgenticFleet routes tasks to specialized AI agents and orchestrates their execution:

Task --> Analysis --> Routing --> Agent Execution --> Quality Check --> Output

Agents: Researcher (web search), Analyst (data/code), Writer, Reviewer, Coder, Planner

Execution Modes:

Mode	Description
Auto	DSPy picks best mode (default)
Delegated	Single agent handles task
Sequential	Agents work in pipeline
Parallel	Concurrent execution
Handoff	Direct agent-to-agent transfers
Discussion	Multi-agent group chat

Usage

CLI

agentic-fleet                                    # Interactive console
agentic-fleet run -m "Your task" --verbose       # Single task
agentic-fleet run -m "Query" --mode handoff      # Specific mode
agentic-fleet list-agents                        # Show available agents

Python API

import asyncio
from agentic_fleet.workflows import create_supervisor_workflow

async def main():
    workflow = await create_supervisor_workflow()
    result = await workflow.run("Summarize transformer architecture")
    print(result["result"])

asyncio.run(main())

Backend API

./scripts/start_backend.sh  # http://localhost:8000
# Docs: http://localhost:8000/api/docs

Configuration

Environment (.env):

OPENAI_API_KEY=sk-...          # Required
TAVILY_API_KEY=tvly-...        # Optional: web search
DSPY_COMPILE=true              # DSPy optimization

Workflow (src/agentic_fleet/config/workflow_config.yaml):

• Models, temperatures, agent settings
• Execution thresholds and limits
• Tracing and evaluation options

Project Structure

src/agentic_fleet/
  agents/        # Agent definitions
  workflows/     # Orchestration logic
  dspy_modules/  # DSPy signatures & reasoner
  tools/         # Tavily, browser, code interpreter
  cli/           # Typer CLI
  app/           # FastAPI backend
src/frontend/    # React UI (optional)
scripts/         # Utilities
docs/            # Documentation

Development

make install           # Install dependencies
make dev               # Run backend + frontend
make test              # Run tests
make check             # Lint + type-check

Documentation

• Getting Started
• Configuration
• Architecture
• Troubleshooting

License

MIT - see LICENSE

Acknowledgments

Built with Microsoft agent-framework, DSPy, and Tavily