⚠️Install with caution. This skill has very few installs. Always review the source and verify it on clawhub.ai before installing. Community-built skills run with agent permissions — only install ones you trust.

⚙️ AI Tools & Utilities

claw-codev1.0.0

Name: claw-code
Author: dae-xiaoe

claw-code

dae-xiaoe

TypeScript runtime port of Claude Code's AI agent harness for running commands, routing prompts, managing sessions, and auditing tool/command registries.

latest

Download Package View on ClawHub

Installs (all time)

Installs (current)

Downloads

672

Stars

CreatedApr 8, 2026

UpdatedMay 11, 2026

Install & Quick Start

Install via ClawdBot CLI:

clawdbot install dae-xiaoe/claw-code

Skill Package59 files

📋SKILL.mdmarkdown

Failed to load file.

Quality Score

B60/100

Grade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.

Market Validation7/35

· 672 downloads (moderate demand)
· No tracked installs (may still have real users via manual install)

Documentation20/25

· SKILL.md present
· Detailed documentation (≥3000 chars)
· Contains usage examples or trigger description
· Detailed summary

Package Completeness8/15

· skillAssets present (58 files)

💡

Usage Guide

Generated Oct 8, 2026

AI engineers and agent framework developersDevOps and platform engineers running agent workloadsTechnical founders building AI assistant productsComputer science researchers and educatorsadvanced

💡 Application Scenarios

Agent Harness Migration ValidationDeveloper Tools / Software Engineering

Engineering teams porting Claude Code's harness from Python or another language use claw-code's parity-audit, command-graph, and tool-pool commands to verify feature coverage against the original TypeScript archive. The turn-loop and route commands lets them reproduce behavior end-to-end before committing to the port.

Prompt Routing Debugging for AI ProductsArtificial Intelligence / SaaS

AI product teams debug why user prompts are mapped to the wrong tools or commands by running route, bootstrap, and turn-loop against representative prompts. They can inspect registry results with commands/tools queries and refine routing rules before shipping to production.

Enterprise Agent Session BootstrappingEnterprise IT / Operations

Platform teams building internal agent assistants use claw-code to bootstrap sessions, persist transcripts with flush-transcript, and reload them via load-session for auditability. This enables repeatable, traceable agent runs across departments such as support, QA, and research.

Agent Harness Education and ResearchEducation / Research

Researchers and educators studying Claude Code internals use summary, manifest, command-graph, and bootstrap-graph outputs as teaching material. Students can explore registries, run route/turn-loop experiments, and compare results against documented behavior.

Remote and Distributed Agent ExecutionCloud Infrastructure / DevOps

Teams running agents across remote machines or containers use remote-mode, ssh-mode, teleport-mode, and direct-connect-mode to execute turns on distributed targets. This suits CI/CD pipelines or edge deployments where central coordination but remote execution is required.

💼 Business Models

Open-Source Harness with Commercial SupportSupport contracts, migration consulting, sponsored feature development

Release claw-code under a permissive license while selling enterprise support, parity audits, and custom registry integrations. The free runtime drives adoption, and paid engagements address migration and compliance needs.

Managed Agent Runtime PlatformUsage-based API pricing, monthly platform subscriptions, enterprise SLAs

Offer a hosted claw-code service where customers send prompts and receive routed, session-managed results via API. The platform handles scaling, transcript persistence, and remote-mode execution so teams avoid self-hosting overhead.

Enterprise Integration and TrainingIntegration project fees, annual training licenses, custom registry development

Package claw-code with domain-specific command and tool registries, delivered alongside workshops on prompt routing and harness architecture. Customers get tailored integrations plus training for their engineering and support teams.

💬 Integration Tip

Run parity-audit and summary first to validate the compiled registry against your target harness, then wrap route and turn-loop behind a thin service layer to control model calls and transcript storage. Keep reference_data snapshots versioned alongside your code so registry updates are reviewable.