Build LangGraph agents like Next.js apps. Dawn is a TypeScript meta-framework that simplifies the development of AI agents and workflows by adopting a filesystem-routing paradigm. It allows developers to define agents, state schemas, tools, and middleware colocated with their respective routes, significantly reducing boilerplate code and enabling engineers to focus on core AI logic. This accelerates the development of complex agent applications for developers familiar with modern web frameworks.

How It Works

Dawn transforms filesystem routes (under src/app/) into runnable AI agents or workflows. It automatically discovers these routes, wires route-local tools into generated LangGraph graphs, and emits a langgraph.json package suitable for deployment and integration with tools like LangSmith. A key component is the dawn dev command, which provides a local development server with Hot Module Replacement (HMR) and Agent Protocol endpoints, facilitating a rapid iteration cycle. The framework also enforces end-to-end type safety by generating TypeScript types for all relevant components, from route parameters to tool inputs/outputs.

Quick Start & Requirements

• Primary install: npm create dawn-ai-app@latest my-dawn-app
• Prerequisites: Node.js 22.12 or later.
• Resources: Official documentation and community support are available via GitHub Discussions.

Highlighted Details

• Filesystem-routed agents: Colocate state schemas, tools, middleware, and tests directly with their corresponding routes for better organization.
• Local Dev Loop: dawn dev provides an HMR development server with Agent Protocol endpoints for rapid, interactive development.
• End-to-End Typing: Generates TypeScript types for route parameters, state, and tool I/O, enhancing code reliability.
• Durable Threads: Built-in SQLite checkpointer and thread store ensure agent state persistence and seamless resumption.
• Integrated Testing: @dawn-ai/testing offers CI-safe harnesses and fixture replay; dawn eval for colocated evaluations.
• Sandboxing: Supports isolating workspace filesystem and shell calls via configurable providers, including a Docker reference implementation.

Maintenance & Community

The project encourages community involvement through GitHub Discussions. Contribution guidelines are detailed in CONTRIBUTING.md, and standards are outlined in AGENTS.md. Security practices are documented in SECURITY.md, and a Code of Conduct is in place.

Licensing & Compatibility

The project is licensed under the MIT License, which is permissive and generally compatible with commercial use and closed-source applications.

Limitations & Caveats

The provided README does not explicitly detail limitations, alpha status, known bugs, or unsupported platforms. The focus is on the framework's capabilities and setup process.