clawdotnet/openclaw.net

Self-hosted Personal AI + agent runtime in .NET (NativeAOT-friendly)

What it solves

OpenClaw.NET provides a self-hosted, NativeAOT-friendly AI agent runtime and gateway specifically for .NET developers. It bridges the gap between high-level agent orchestration and low-level system deployment, offering a path from source code to high-performance release artifacts while maintaining compatibility with the OpenClaw ecosystem.

How it works

The project implements a runtime loop that manages tool execution, streaming, and session memory. It includes a gateway that exposes OpenAI-compatible HTTP endpoints, a chat UI, and an admin interface. The system supports multiple LLM providers (OpenAI, Claude, Gemini, DeepSeek, Ollama) and can even run embedded local models like Gemma 4 via GGUF packages. It also integrates with the Model Context Protocol (MCP) for third-party app discovery and tool bridging.

Who it’s for

.NET developers and operators who need a local or self-hosted agent gateway with explicit diagnostics, first-party .NET tools, and the ability to deploy as NativeAOT binaries.

Highlights

  • NativeAOT-friendly: Optimized for high-performance .NET deployment.
  • Extensive Tooling: Over 80 native tool surfaces and 9 channel adapters (e.g., Telegram, Slack, Discord).
  • Governance & Observability: Includes a passive governance ledger, evidence bundles for human review, and a dedicated admin UI for trajectory and audit exports.
  • Local Model Support: Built-in support for Ollama and embedded Gemma 4 models with frame-based video understanding.
  • Agentic Capabilities: Supports session-scoped goal auto-continuation (/goal) and recurring-prompt commands (/loop).