claw-code for Openclaw

A TypeScript port of the Claude Code harness system designed for AI agent execution and runtime auditing.

dae-xiaoe
v1.0.0
Apr 8, 2026
0
711
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install claw-code

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install claw-code using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is claw-code?

claw-code is a specialized runtime and harness system ported to TypeScript, specifically designed to mirror the internals of Claude Code. As one of the core Openclaw Skills, it provides a robust framework for bootstrapping agent sessions, routing prompts to specific tools or commands, and maintaining parity with original TypeScript archives. It serves as a bridge for developers who need to audit agent behavior, query tool registries, or execute complex turn loops in a controlled environment.

By leveraging this skill, developers can interact with a mirrored version of the Claude Code ecosystem, allowing for deep inspection of subsystems, command graphs, and tool pools without leaving their primary development environment.

claw-code Use Cases

  • Investigating Claude Code internals and harness runtime behavior.
  • Routing user prompts to specific mirrored commands or tools within the Openclaw Skills ecosystem.
  • Bootstrapping agent sessions to handle complex multi-turn interactions.
  • Performing parity audits against original TypeScript code archives to ensure consistency.
  • Executing remote or direct connect modes for specialized agent communication scenarios.

How claw-code Works

  1. The user initiates a command through the CLI entry point located in the scripts directory.
  2. The Runtime Core loads necessary tool and command registries from the provided reference data snapshots.
  3. The Query Engine processes user prompts to determine the appropriate routing logic or bootstrap stage.
  4. The agent executes the specified turn loop or command, interacting with the unified Tool Pool.
  5. Session Management components persist the transcript and history to ensure session continuity and auditability.

claw-code Setup

To get started with this entry in the Openclaw Skills library, navigate to the scripts directory and run the compiled entry point:

cd .openclaw/workspace/skills/claw-code/scripts
node dist/main.js summary

If you prefer to build the harness from source:

cd .openclaw/workspace/skills/claw-code/scripts
npm install
npm run build

claw-code Data Schema & Taxonomy

The skill organizes its internal data and reference material using a structured filesystem approach:

Component Description
reference_data/ Contains JSON snapshots of tools, commands, and subsystems used by the runtime.
dist/main.js The primary compiled execution entry point for CLI operations.
src/ The TypeScript source files including core runtime logic and graph definitions.
session_store.ts Manages the persistence and retrieval of active agent session states.

claw-code Advanced Features

  • Multi-mode remote execution support including SSH, Teleport, and Deep Link modes.
  • Comprehensive parity auditing tools to compare current runtime behavior against historical archives.
  • Visual command graph and bootstrap stage reporting for debugging complex agent logic.
  • High-performance query engine with integrated streaming support for prompt routing.
  • Flexible tool management allowing users to exclude MCP tools or toggle simple mode execution.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*