Welcome to AZM AI, an advanced platform for software development agents powered by AI.

AZM AI agents can do anything a human developer can: modify code, run commands, browse the web, call APIs, and yes—even copy code snippets from StackOverflow.

Learn more in our GitHub Wiki to get started.

[!IMPORTANT] Using AZM AI for work? We'd love to chat! Contact us at Steve6546 to learn more about our services and how AZM AI can help your business.

☁️ AZM AI Cloud

The easiest way to get started with AZM AI is on our cloud platform, contact us to learn more about our cloud offerings.

💻 Running AZM AI Locally

AZM AI can also run on your local system using Docker. See our Wiki for system requirements and more information.

[!WARNING] On a public network? See our Security Guide to secure your deployment by restricting network binding and implementing additional security measures.

```bash docker pull ghcr.io/steve6546/azm-ai/runtime:latest

docker run -it --rm --pull=always \ -e SANDBOXRUNTIMECONTAINERIMAGE=ghcr.io/steve6546/azm-ai/runtime:latest \ -e LOGALL_EVENTS=true \ -v /var/run/docker.sock:/var/run/docker.sock \ -v ~/.azm-ai-state:/.azm-ai-state \ -p 3000:3000 \ -p 12001:12001 \ --add-host host.docker.internal:host-gateway \ --name azm-ai-app \ ghcr.io/steve6546/azm-ai/azm-ai:latest ```

You'll find AZM AI running at http://localhost:3000!

When you open the application, you'll be asked to choose an LLM provider and add an API key. Anthropic's Claude 3.5 Sonnet (anthropic/claude-3-5-sonnet-20241022) works best, but you have many options available.

💡 Other ways to run AZM AI

[!CAUTION] AZM AI is meant to be run by a single user on their local workstation. It is not appropriate for multi-tenant deployments where multiple users share the same instance. There is no built-in authentication, isolation, or scalability.

If you're interested in running AZM AI in a multi-tenant environment, please contact us for advanced deployment options.

You can also connect AZM AI to your local filesystem, run AZM AI in a scriptable headless mode, interact with it via a friendly CLI, or run it on tagged issues with a github action.

Visit our Wiki for more information and setup instructions.

If you want to modify the AZM AI source code, check out Development.md.

Having issues? Check our Troubleshooting Guide in the Wiki.

📖 Documentation

To learn more about the project, and for tips on using AZM AI, check out our documentation.

There you'll find resources on how to use different LLM providers, troubleshooting resources, and advanced configuration options.

🤝 How to Join the Community

AZM AI is a community-driven project, and we welcome contributions from everyone. We do most of our communication through GitHub, so this is the best place to start:

• Read or post Github Issues - Check out the issues we're working on, or add your own ideas.
• Contact us directly - For business inquiries or partnership opportunities.

See more about the community in COMMUNITY.md or find details on contributing in CONTRIBUTING.md.

📈 Progress

See the AZM AI roadmap here for our upcoming features and improvements.

📜 License

Distributed under the MIT License. See LICENSE for more information.

🙏 Acknowledgements

AZM AI is built by a dedicated team of contributors, and every contribution is greatly appreciated!

AZM AI is an advanced AI development platform built under the MIT license. We've created a comprehensive solution with innovative features and capabilities to enhance AI-driven development workflows.

For a list of open source projects and licenses used in AZM AI, please see our CREDITS.md file.

📚 References

AZM AI is based on advanced AI research and development in the field of generalist agents for software development.

🏗️ Project Structure

AZM AI is organized into several key components:

Core Components

• azm_ai/core: Core functionality and configuration
• azm_ai/events: Event handling system for actions and observations
• azm_ai/runtime: Runtime environments for executing agent actions
  • impl/modal: High-availability runtime implementation
  • impl/e2b: File system operations implementation
  • impl/browsergym: Web browsing capabilities
• azm_ai/server: Server implementation including API endpoints and middleware
• azm_ai/storage: Data persistence layer
• azm_ai/utils: Utility functions used throughout the codebase

Frontend

• frontend: React-based user interface
  • src/components: UI components
  • src/pages: Application pages
  • src/hooks: Custom React hooks
  • src/utils: Frontend utility functions

Testing

• tests: Comprehensive test suite
  • unit: Unit tests for individual components
  • integration: Integration tests for component interactions
  • e2e: End-to-end tests for complete workflows

Documentation

• docs: Project documentation
  • static: Static assets including images
  • api: API documentation
  • guides: User and developer guides

🧩 Architecture

AZM AI follows a modular architecture with clear separation of concerns:

1. Event-Driven Core: The system is built around an event stream that handles actions and observations
2. Pluggable Runtimes: Different runtime environments can be used for executing actions
3. Extensible API: The server exposes a RESTful API for interacting with the system
4. Responsive UI: The frontend provides a user-friendly interface for interacting with AI agents

This architecture allows for easy extension and customization of the platform for different use cases.