Claw Code is an open-source AI coding agent framework, offering a clean-room rewrite of the Claude Code architecture. Built entirely from scratch using Python and Rust, it provides a robust platform for AI development without relying on proprietary code or model weights. This innovative framework is designed for multi-agent orchestration, tool-calling, and terminal-native AI development.

Key Features

• Plugin-Based Tool System: Claw Code includes 19 permission-gated tools for file I/O, shell execution, Git operations, and more, each independently sandboxed with configurable access controls.

• Autonomous Agent Loop: The terminal-native agent reads your entire codebase, edits files, executes commands, runs tests, and iterates autonomously until the task is complete.

• Rust Performance Core: A 6-crate Rust workspace with performance-critical paths, zero-dependency JSON parser, and syntax-highlighted terminal rendering.

• LLM API Client: A provider-agnostic API client with automatic retry, SSE streaming, OAuth authentication, and token usage tracking with cost estimation.

Claw Code's architecture is informed by the Claude Code source leak, reimplementing core patterns such as the tool system, query engine, and multi-agent orchestration. The framework supports spawning sub-agents to parallelize complex tasks, enabling decomposition of large engineering challenges into independently executable subtasks. With full Model Context Protocol (MCP) support, Claw Code connects to external tool servers with automatic name normalization, config hashing, and OAuth authentication.

Explore Claw Code on GitHub to dive deeper into its architecture, tools, and deployment strategies. As an open-source project, it welcomes community contributions and offers a comprehensive platform for AI coding agent development.