RayClaw

English | 中文

Note: This project is under active development. Features may change, and contributions are welcome!

RayClaw is a multi-channel agentic AI runtime written in Rust. It connects to Telegram, Discord, Slack, Feishu/Lark, and a built-in Web UI through a unified agent engine. Every conversation flows through the same tool-calling loop — shell commands, file operations, web search, background scheduling, and persistent memory — regardless of which channel it arrives on. RayClaw works with multiple LLM providers (Anthropic + any OpenAI-compatible API) and is designed for extensibility: adding a new channel adapter requires no changes to the core agent engine.

Table of contents

• How it works
• Install
• Desktop App
• Use as Rust crate
• Features
• Tools
• Memory
• Skills
• MCP
• ACP (Agent Client Protocol)
• Plan & Execute
• Scheduling
• Local Web UI (cross-channel history)
• Release
• Setup
• Configuration
• Platform behavior
• Multi-chat permission model
• Usage examples
• Architecture
• Adding a New Platform Adapter
• Documentation

Install

One-line installer (recommended)

curl -fsSL https://rayclaw.ai/install.sh | bash

Preflight diagnostics

Run cross-platform diagnostics before first start (or when troubleshooting):

rayclaw doctor

Machine-readable output for support tickets:

rayclaw doctor --json

Checks include PATH, shell runtime, Node/npm, agent-browser, and MCP command dependencies from rayclaw.data/mcp.json.

Uninstall (script)

curl -fsSL https://rayclaw.ai/uninstall.sh | bash

From source

git clone https://github.com/rayclaw/rayclaw.git
cd rayclaw
cargo build --release --features all
cp target/release/rayclaw /usr/local/bin/