Zao Skill for Openclaw

Zao Skill is a meta-capability framework designed to help developer agents self-improve, design, and package high-efficiency Openclaw Skills following strict prompt engineering and progressive disclosure principles.

sensenkawa
v0.8.5
Jun 2, 2026
0
530
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install zao-skill

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 zao-skill 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 Zao Skill?

Zao Skill is a meta-engineering utility designed to orchestrate the lifecycle of agent capabilities. It empowers developers and AI coding assistants to design, draft, validate, and package highly optimized Openclaw Skills. By enforcing core principles like absolute conciseness and progressive disclosure, the skill ensures that agent instruction sets remain lightweight, context-efficient, and structurally sound.

Instead of overloading an AI's context window with verbose guidelines, this skill provides a rigorous methodology for splitting instructions into tiered layers: metadata, core SKILL.md workflows, and external references. Using Zao Skill, developers can automate static validation, perform cognitive bias checks, and seamlessly package skills for distribution, dramatically improving agent efficiency and accuracy in execution.

Zao Skill Use Cases

  • You need to design a structured, reliable capability for an AI assistant or coding agent from scratch.
  • You want to refactor an existing monolithic instruction set into a highly efficient, multi-tiered architecture using Openclaw Skills.
  • You want to enforce quality gates, static validation, and package compiled skills (.skill files) for deployment.
  • Your AI agents suffer from context window saturation, and you need a system to prune and optimize prompt tokens.
  • You are tracking and fixing recurring agent failures, edge cases, or cognitive laziness across multiple workflows.

How Zao Skill Works

  1. Pre-Step Rationalization: Before executing any action, the agent performs a mandatory bias check (re-reading files, verifying references, and assessing completion) to ensure it acts on actual disk files, not stale memory state.
  2. Workflow Extraction (Phase 1): The skill analyzes user intent, extracts a clear procedural workflow, and records a draft definition to a temporary .wip file.
  3. Similar Skill Analysis: It conducts local or online searches to compare and enhance the design using existing industry-standard Openclaw Skills.
  4. Structured Drafting (Phase 2): It generates the directory structure and populates SKILL.md with required frontmatter, core directives, step-by-step workflows, and exit gates.
  5. Static and Checklist Validation (Phase 3): Zao Skill executes a validator script and steps through a comprehensive quality checklist, capturing evidence to ensure zero-fail execution.
  6. Meta Evolution & Packaging (Phase 4): On success, the framework compiles the directory into a production-ready capability and appends discovered gotchas to the evolution table.

Zao Skill Setup

Integrate Zao Skill into your workspace by setting up the directory structure and configuring the essential validation scripts.

# Create a new skill directory structured for progressive disclosure
mkdir -p my-custom-skill/.wip
mkdir -p my-custom-skill/references
mkdir -p my-custom-skill/scripts

# Run the validation script to check compliance with Openclaw Skills standards
python3 scripts/quick_validate.py ./my-custom-skill

# Compile and package your finalized skill directory into a distributable format
python3 scripts/package_skill.py ./my-custom-skill ./dist

Zao Skill Data Schema & Taxonomy

Zao Skill implements a strict, modular taxonomy to support progressive disclosure and ensure efficient token consumption for Openclaw Skills.

Component Path Description Required
Frontmatter SKILL.md (Top) YAML metadata including kebab-case name and trigger-focused description. Yes
Instruction Body SKILL.md Core operational directives, step-by-step workflow, and exit verification checklist. Yes
Work in Progress .wip/ Houses temporary workflow extractions and intermediate validation summaries. No
References references/ Domain-specific Markdown files containing complex logic loaded on-demand. No
Scripts scripts/ Executable scripts (e.g., Python validators, packagers) for deterministic processing. No
Assets assets/ Static files, icons, templates, and font configurations used in outputs. No

Zao Skill Advanced Features

  • Meta Evolution Check: Seamlessly logs execution gotchas, successful patterns, and process improvements back into target Openclaw Skills dynamically post-run.
  • Pre-Step Rationalization Check: Overcomes model cognitive laziness and memory bias by enforcing fresh disk-reads and deep reference-file lookups before every operation.
  • Progressive Disclosure Loading: A smart three-tiered architecture that selectively exposes detailed prompt instructions to the LLM context only when triggered.
  • Deterministic Packaging Automation: Built-in Python scripts that execute compliance checks and build production-ready files in a single CLI command.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*