About Agent-Native

Open-source apps for agents and people

Agent-Native is an open-source framework for building applications where AI agents and user interfaces share the same actions, data, and application state.

A shared operating model

Agent-Native treats the agent and the interface as equal partners. A single action can power a UI control, an agent tool, an HTTP endpoint, an MCP or A2A capability, a CLI command, and an auditable workflow. Shared SQL state keeps the human view and the agent view aligned instead of creating a second hidden system.

Open source by default

The framework source is available under the MIT license in the BuilderIO/agent-native repository. Developers can inspect the implementation, run it locally, choose their own database and model providers, and adapt the framework to the needs of their product. Hosted Agent-Native services are operated separately from forks and self-hosted deployments.

Hosted and self-hosted

Builder.io operates the hosted Agent-Native applications and documentation at agent-native.com. The framework is also designed for teams that want to deploy and support their own applications. The same action-first contracts, access boundaries, agent instructions, and public protocols can be reviewed in the source and docs.

Built in public

The project is developed in the open through GitHub issues, pull requests, documentation, and the Agent-Native community. Read the docs to learn the architecture, browse the source to verify an implementation, or join the community when you want to discuss a use case or contribute a change.