A specialized meta-skill designed to streamline the creation, modification, evaluation, and packaging of robust Openclaw Skills.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install dazhuangskill-creator
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install dazhuangskill-creator 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 Dazhuangskill Creator?
Dazhuangskill Creator is a professional framework built to manage the entire lifecycle of AI coding agent capabilities. It treats the skill directory as a modular base, enforcing a lean philosophy that separates core logic from secondary resources. This ensures that the resulting Openclaw Skills are efficient, maintainable, and highly effective for complex developer workflows.
By prioritizing a hierarchical structure, this skill helps developers move away from bloated single-file prompts into organized packages. It leverages a state-managed workflow to ensure the AI agent remains focused during multi-step development tasks, making it an essential tool for anyone building within the ecosystem of Openclaw Skills.
Dazhuangskill Creator Use Cases
- Transforming complex manual coding workflows into repeatable Openclaw Skills.
- Refactoring and modularizing existing agent instructions to reduce context bloat.
- Benchmarking and validating the quality of agent outputs using standardized evaluation loops.
- Optimizing skill trigger descriptions to improve activation accuracy in multi-agent environments.
- Packaging finalized skills for delivery with structured metadata and environment-specific configurations.
How Dazhuangskill Creator Works
- Task Identification: The skill first determines if the user path is to create, modify, evaluate, optimize, or package a skill.
- Structural Planning: It defines the minimum required information and maps out which components belong in the main body versus external references.
- Scaffolding and Drafting: Utilizing internal scripts, it generates the necessary file structure and drafts the core logic following a lean-first approach.
- Validation Path: It selects the lightest yet most effective verification method, ranging from sanity checks to blind A/B comparisons.
- Iterative Refinement: The skill summarizes changes, prunes unnecessary metadata, and determines if the output meets the quality standards for Openclaw Skills.
Dazhuangskill Creator Setup
To begin using Dazhuangskill Creator, navigate to your skill base directory and initialize your environment. Ensure you have Python 3 installed to run the supporting automation scripts provided for Openclaw Skills.
# Navigate to your skill directory
cd "<skill-base>"
# Initialize a new skill structure
python3 scripts/init_skill.py --name your-skill-name
# Perform a quick validation of an existing skill
python3 scripts/quick_validate.py <skill-dir>
Dazhuangskill Creator Data Schema & Taxonomy
Dazhuangskill Creator organizes Openclaw Skills using a strict file taxonomy to ensure portability and clarity:
| File/Directory | Role | Description |
|---|---|---|
SKILL.md |
Main Logic | Contains durable rules, default workflows, and internal pointers. |
references/ |
Documentation | Stores schemas, long explanations, and low-frequency modules. |
assets/ |
Templates | Static files and templates that the agent copies or fills during execution. |
scripts/ |
Automation | Deterministic Python or Bash scripts for repetitive tasks. |
config.yaml |
Parameters | Human-readable configuration for frequent parameter adjustments. |
agents/ |
Sub-Agents | Specialized instructions for graders, comparators, or analyzers. |
Dazhuangskill Creator Advanced Features
- State-aware workflow tracking using current_path and current_step to maintain context in long conversations.
- Automated generation of OpenAI-compatible metadata via generate_openai_yaml.py for seamless integration.
- Support for diverse runtime environments including Claude.ai, Codex, and Cowork via specialized reference guides.
- Built-in evaluation logic for Openclaw Skills that includes blind A/B testing and automated grading.
- Conventional Commit enforcement for skills that generate high-density, structured outputs like changelogs or PR titles.
SKILL.md
Loading
Related Openclaw Skills
OrderKeeper
A post-purchase operations skill that manages order lifecycles, after-sales triage, and consumer protection deadlines.
PriceTide
PriceTide is a decisive shopping timing strategist that determines if a current price is a short-term low or if you should wait for better promotion windows.
LinkedIn B2B Buying Signal Detector
An AI-driven sales intelligence tool that monitors LinkedIn hiring signals and growth patterns to identify and score high-intent B2B leads.
High-Intent Lead Sniper
A high-performance sales intelligence skill that detects real-world buying signals to generate hyper-personalized outreach for prospects ready to purchase.
Lobster Attachment Inject
A dynamic content injection system that manages agent skills and metadata as attachments to reduce token overhead and preserve prompt caching.
ReviewLens
ReviewLens is a decision-intelligence layer that transforms noisy marketplace comments into structured buying judgments and risk assessments.
Featured*
