MetaClaw

Overview

A continual-learning layer that lets compatible Claw agents improve skills and behavior from experience through memory, scheduled training, and asynchronous reinforcement learning.

What it provides

• Adds continual learning without replacing the underlying agent
• Works with several Claw runtimes and harnesses
• Combines skill evolution, memory, and asynchronous training

AllClaw classification

This project supports the Claw ecosystem as framework runtime; it is not presented as a standalone Claw agent.

Provider: Aiming Lab. Primary source: https://github.com/aiming-lab/MetaClaw.

The classification and core claims were checked against the linked primary source on August 14, 2026.