Nm Abstract Modular Skills

Build composable skill modules with hub-and-spoke loading

This is a documentation-only skill for designing modular skills; it does not request secrets or install code. Before relying on its command examples, confirm the referenced scripts (skill-analyzer, token-estimator, module_validator, etc.) actually exist in your environment or repository — the SKILL.md assumes those tools are available elsewhere. Also be aware the enforcement-patterns module recommends strong, prescriptive trigger language (e.g., 'YOU MUST'), which can make an agent prefer this skill inappropriately; review and soften enforcement language if you want more balanced behavior. If you plan to run any scripts the documentation references, inspect those script files first (in the source repo or plugin) so you know what filesystem or network access they perform.

Type: OpenClaw Skill Name: nm-abstract-modular-skills Version: 1.8.3 The skill bundle provides a comprehensive framework and documentation for designing modular AI agent skills, focusing on token optimization and progressive disclosure. While SKILL.md and various modules (e.g., core-workflow.md, troubleshooting.md) instruct the agent to execute shell commands and Python scripts (such as scripts/analyze.py and module_validator), these actions are contextually appropriate for the stated purpose of skill analysis and validation. No evidence of data exfiltration, malicious code, or harmful prompt injection was found; the 'enforcement language' patterns are templates for skill steering rather than attacks on the agent's security boundaries.