This project outlines the core principles of building with the OpenAI Agents SDK using a cloud-first, agent-native approach. DACA stands for a development framework where AI is at the center of application logic, with a focus on flexibility, portability, and reliability.
-
AI-First Approach
Everything is driven by agentic AI — decisions, logic, and interactions are handled through agents. -
Agent-Native, Cloud-First
Our agents are designed to run seamlessly across cloud-native environments. -
Develop Anywhere
You can build, test, and deploy your agents from any development environment. -
Cloud Anywhere
Agents are designed to run on any cloud provider — AWS, GCP, Azure, etc. -
Open Core with Managed Edges
All components are open source, with managed standards ensuring reliability. We avoid unknown or unstable packages that cause production issues or developer delays.
The LLM (Large Language Model) supports the following 4 elements to function effectively:
-
System Prompt / Persona
Defines the role, behavior, and tone of the agent. -
User Prompt
The actual input/query given by the user. -
Tool's Schema
Defines how tools should be structured and understood by the agent. -
Tool's Message
The response or action performed by the tool, sent back to the LLM.
- OpenAI Agents SDK
- Python 3.11+
- ASGI Frameworks
- Cloud-Native Dev Environments
To enable the development of intelligent, tool-using agents that are cloud-flexible, scalable, and built on open standards — simplifying complex orchestration with clean, maintainable code.
Feel free to fork, improve, and suggest enhancements via pull requests. Let’s build the future of agentic AI — together.
MIT — use freely with attribution.