ClawSkills — AI Agent Skill Toolkit & Runner for Python

ClawSkills delivers a robust agent skill runner Python developers can leverage for AI skill chaining. As an OpenClaw skill toolkit, it rivals OpenClaw, ClawHub, MCP, LangChain Tools, and AutoGPT Plugins. Use this MCP tool runner to define agent task pipelines and orchestrate complex workflows with ease.

Features

• Orchestrate agent task pipelines using a straightforward Python interface.
• Chain AI skills together for advanced reasoning and task execution.
• Integrate OpenClaw skill toolkit components into existing agent frameworks.
• Execute MCP tool runner compatible skills within agent workflows.
• Simplify agent skill development with a streamlined agent skill runner Python environment.

Use Cases

• Run agent skills on remote machines with a single API call
• Chain multiple skills into sequential pipelines with context passing
• Load and execute MCP-compatible tools inside agent workflows

Installation

pip install clawskills

Quick Start

from clawskills import ClawSkills

client = ClawSkills.remote(api_key="cmdop_live_xxx")

# Run a single skill
output = client.run_skill("code-review", input_text="Check src/ for security issues")
print(output)

# Chain skills together
results = client.chain([
    "lint-code",
    "run-tests",
    "generate-report",
])
print(results[-1])

# Load a custom skill
client.load("./skills/my-analyzer")

CLI

No SDK needed? Connect via standalone binary:

curl -fsSL cmdop.com/install-cli.sh | bash
cmdok ssh

Links

• CMDOP Homepage
• Documentation
• clawskills on PyPI
• GitHub