DotHarness/dotcraft

What it solves

DotCraft provides a project-native AI agent runtime that turns software projects into extensible environments where AI agents can operate. It solves the problem of fragmented agent state by ensuring that conversations, memory, agents, and skills remain tied to the project itself, allowing users to switch entry points without losing progress.

How it works

Built with C#/.NET, DotCraft acts as an embeddable runtime that can be deployed as a desktop application, a server with an SDK, or integrated directly into existing .NET applications. It utilizes trajectory tracking to maximize prefix cache reuse, which helps reduce token costs across different sessions. It supports multiple model providers, including OpenAI, Anthropic, and ChatGPT subscriptions.

Who it’s for

It is designed for .NET developers and teams who want to integrate advanced agent capabilities—such as planning, subagents, and dynamic workflows—into their own desktop tools, services, or automation projects.

Highlights

• Built-in Agent Capabilities: Includes support for planning, agent teams, goals, dreams, and automations.
• Project-Native State: Memory and configurations travel with the project folder.
• Flexible Deployment: Can be run locally, on a private server, or embedded into other apps via APIs and SDKs.
• Cost Optimization: Uses trajectory tracking to lower token usage via prefix cache reuse.