Summary

OpenClaw.NET addresses the gap for .NET developers needing a self-hosted AI agent runtime and gateway. It provides a .NET-native alternative to Python/Node.js stacks, enabling lean, self-contained binary deployments via NativeAOT and facilitating the reuse of existing OpenClaw ecosystem assets. The project targets .NET engineers building production-grade AI agent infrastructure, offering broad LLM integration, extensive tooling, and multi-channel communication capabilities.

How It Works

This project implements a .NET-based agent runtime and gateway, prioritizing NativeAOT compatibility for efficient deployments. Its core features include a multi-step execution engine with robust tool calling, retry mechanisms, per-call timeouts, and circuit-breaker patterns. It supports practical reuse of OpenClaw TS/JS plugins and SKILL.md packages, abstracting LLM interactions through provider-agnostic model profiles. The architecture emphasizes a flexible message pipeline, session management, and a tool execution layer with policy enforcement and routing.

Quick Start & Requirements

• Primary install/run: Clone the repository (git clone https://github.com/clawdotnet/openclaw.net), navigate into the directory, and run dotnet run --project src/OpenClaw.Cli -c Release -- setup. Alternatively, if the CLI is in the PATH, use openclaw setup.
• Prerequisites: .NET SDK, Git. API keys for LLM providers (e.g., MODEL_PROVIDER_KEY environment variable) are required for specific provider configurations. Docker deployment is also supported.
• Links:
  • Repository: https://github.com/clawdotnet/openclaw.net
  • Documentation: Quickstart Guide, User Guide, Compatibility Guide, Security Guide, and others are linked within the README.

Highlighted Details

• Tooling: Includes 48 native tools covering file operations, sessions, memory management, web search, messaging, home automation, databases, and more.
• Channels: Integrates 9 channel adapters (Telegram, SMS, WhatsApp, Teams, Slack, Discord, Signal, Email, Webhooks) with configurable policies like DM handling and sender allowlists.
• LLM Support: Native support for OpenAI, Claude, Gemini, Azure OpenAI, Ollama, and OpenAI-compatible endpoints, with provider-agnostic model profiles.
• NativeAOT: Optimized for NativeAOT compilation, enabling trim-safe, low-memory, self-contained executable deployments.
• Evolving Workflows: Features review-first self-evolving capabilities that propose profile updates, automation drafts, and skill drafts based on session history.
• Plugin System: Supports JS/TS bridge plugins, native dynamic .NET plugins (JIT mode), and standalone SKILL.md packages, with installation and review mechanisms.

Maintenance & Community

Information regarding specific maintainers, community channels (like Discord/Slack), or a public roadmap is not detailed in the provided text. The project encourages contributions via GitHub issues.

Licensing & Compatibility

The specific open-source license for this project is not explicitly stated in the provided README excerpt. It is presented as an independent implementation inspired by OpenClaw. Compatibility is emphasized with the OpenClaw TS/JS plugin ecosystem and SKILL.md packages. Integration with Semantic Kernel is also supported.

Limitations & Caveats

This project is an independent implementation and is not affiliated with the original OpenClaw. The NativeAOT runtime mode (aot) has limitations, supporting only native tools and bridge plugins, excluding full JIT plugin surfaces. The specific software license is not disclosed in the provided documentation.