Skip to content

Universal AI agent deployment platform - Deploy any agent to any cloud with a single command.

Notifications You must be signed in to change notification settings

zenml-io/panagent

Repository files navigation

PanAgent

Universal AI agent deployment platform - Deploy any agent to any cloud with a single command.

PanAgent lets you package any LLM-powered agent into a FastAPI Docker container and deploy it anywhere – local dev, Google Cloud Run, AWS Fargate, AWS Lambda – with a single CLI.

✨ Key Features

  • One-command panagent deploy workflow
  • Adapter system for PydanticAI, LangGraph, LangChain, LlamaIndex, Haystack, Google ADK, OpenAI Agents SDK, AWS Strands SDK, AutoGen, CrewAI & Semantic Kernel agents
  • Automatic port management for local dev
  • Container images with fully-pinned Python requirements for reproducible builds
  • Deployment Registry so you can list, update and clean-up services safely
  • Works with Python 3.12+ and Docker on macOS, Linux, Windows

🔗 Full documentation: see the PanAgent Docs for detailed framework guides, deployment tutorials, API & CLI references and the complete panagent.yaml schema.


🚀 Quick Start (Local)

# 1 – Create a project (example: PydanticAI)
panagent init --framework pydanticai --name my-agent
cd my-agent

# 2 – Build the Docker image
panagent build

# 3 – Run it locally (auto-selects a free port)
panagent deploy --target local

# 4 – Call your agent
panagent test --input '{"message": "Hello!"}'

Ready for the cloud? Follow the step-by-step guides for Google Cloud Run, AWS Fargate or AWS Lambda.


📦 Installation

Bare CLI:

pip install panagent

With a framework extra:

pip install "panagent[langgraph]"     # or [pydanticai], [haystack], [adk], [autogen], [openai-agents], [crewai], [llama-index]

Cloud helpers:

pip install "panagent[gcp]"           # Cloud Run tools
pip install "panagent[aws]"           # Fargate & Lambda tools
pip install "panagent[all]"           # Everything

📋 Requirements

  • Python 3.12+
  • Docker
  • OpenAI API key (OPENAI_API_KEY)
  • Cloud CLIs for optional deployments (gcloud, aws)

🏗️ Architecture

┌─────────────────────┐
│   Agent Framework   │
│ (PydanticAI/        │
│  LangGraph/LangChain/│
│  LlamaIndex/        │
│  Haystack/ADK/AutoGen) │
└──────────┬──────────┘
           │
     ┌─────▼─────┐
     │ PanAgent  │
     │  Adapter  │
     └─────┬─────┘
           │
    ┌──────▼──────┐
    │   FastAPI   │
    │  Container  │
    └──────┬──────┘
           │
  ┌────────┴────────┐
  │                 │
┌─▼──┐  ┌────┐  ┌─────┐  ┌──▼───┐
│GCP │  │AWS │  │λ AWS│  │Local │
└────┘  └────┘  └─────┘  └──────┘

🗺️ Roadmap (abridged)

  • ✅ PydanticAI, LangGraph, LangChain, Haystack, Google ADK support
  • ✅ Local, GCP Cloud Run, AWS Fargate, AWS Lambda targets
  • 🔜 OpenTelemetry tracing, advanced auth, performance tuning, access controls for public deployments

🤝 Contributing • 📄 License • 💬 Support

We love contributions! See [CONTRIBUTING.md]. MIT License – see [LICENSE]. Questions? [GitHub Issues] or hello@c.

About

Universal AI agent deployment platform - Deploy any agent to any cloud with a single command.

Resources

Stars

Watchers

Forks

Releases

No releases published

Contributors 2

  •  
  •  

Languages