rsclaw-ai/rsclaw

Registry

Summary

RSCLAW is an AI Agent Engine built in Rust, featuring long-term memory and self-learning capabilities, with one-click migration from OpenClaw.

Key features

• Long-term memory for persistent context
• Self-learning capability for adaptive behavior
• One-click migration from OpenClaw
• Built with Rust for high performance and safety
• Agent engine designed for autonomous tasks

Use cases

• Building autonomous AI agents with memory
• Migrating existing OpenClaw projects seamlessly
• Developing self-improving AI systems
• Creating persistent conversational agents
• Powering adaptive automation workflows

README excerpt

# RsClaw > **An AI agent engine that remembers — and gets better the more you use it.** > One 15MB binary · 13 channels · 15 LLM providers · Multi-backend agents · OpenCLI-ready · Built in pure Rust. [](https://github.com/rsclaw-ai/rsclaw/stargazers) [](https://crates.io/crates/rsclaw) [](https://github.com/rsclaw-ai/rsclaw/releases) [](https://crates.io/crates/rsclaw) [](#license) [](https://www.rust-lang.org/) **🇺🇸 English** · [🇨🇳 中文](docs/lang/README_cn.md) · [🇯🇵 日本語](docs/lang/README_ja.md) · [🇰🇷 한국어](docs/lang/README_ko.md) · [More languages ▾](docs/lang/) <p align="center"> <img src="docs/images/en.gif" alt="RsClaw Preview" width="800" /> </p> Most AI agents forget everything between sessions. Every new conversation starts from zero — your preferences, your context, your workflow, all gone. **RsClaw doesn't forget.** Built from scratch in Rust, RsClaw (Crab AI / 螃蟹 AI) persists every interaction through a three-layer memory store (redb + tantivy + hnsw_rs), learns from your usage patterns, and ships as a single 15MB binary running on ~20MB RAM. Four agent lifetime modes (Main/Named/Sub/Task), cap-protocol coding-agent integration (Claude Code/OpenClaude/OpenCode/Codex via `tool_cap`), 13 messaging channels, 15 LLM providers, A2A cross-machine orchestration — all without a line of Node.js. Drop-in OpenClaw replacement. 💬 [

Topics

No topics yet.