A comprehensive framework for architecting, securing, and publishing professional-grade Openclaw Skills using standardized MLOps and security gatekeeping.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ai-skill-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 ai-skill-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 AI Skill Creator?
The AI Skill Creator is a high-level developer workflow designed to standardize how new capabilities are built for AI agents. It bridges the gap between technical implementation and security compliance by integrating CTO-led MLOps lifecycle phases with CISO-led security reviews. This ensures that all Openclaw Skills are not only functional but also adhere to zero-trust architecture and minimal permission principles.
By providing a structured path from ideation to publishing, this skill automates the creation of necessary directory structures, enforces documentation standards, and runs mandatory security scans. It is specifically designed to output compliant .skill packages ready for review and deployment in production-grade AI environments.
AI Skill Creator Use Cases
- Creating new AI agent capabilities from scratch with standardized directory structures.
- Automating security audits for agent scripts using STRIDE modeling and CVSS scoring.
- Enforcing quality gates and documentation standards before publishing to a skill marketplace.
- Developing inter-agent interfaces that allow isolated and secure task delegation between different AI models.
How AI Skill Creator Works
- Initialization: The workflow starts by executing a script to scaffold a compliant directory structure, including specialized folders for scripts, references, and assets.
- Requirement Analysis: It performs a deep dive into the skill's triggers, core functions, and necessary permissions, generating a formal design document.
- Implementation: Developers author the SKILL.md and supporting logic following strict coding red-lines to prevent vulnerabilities like path traversal or credential leakage.
- Security Review: A mandatory phase evaluates the skill against security benchmarks, requiring authorization for any high-risk operations.
- Quality Gate: The skill must pass several automated and manual checks (G0-G7) covering structure, description quality, and script execution.
- Packaging & Publishing: Once approved, the skill is bundled into a .skill artifact and pushed to the repository for wider use within the ecosystem of Openclaw Skills.
AI Skill Creator Setup
To begin building Openclaw Skills using this workflow, ensure you have the environment configured and use the initialization scripts provided:
# Initialize a new skill directory
python3 ./skills/skill-creator/scripts/init_skill.py <your-skill-name> --path ~/.qclaw/skills
# Run a security review before packaging
python3 ./skills/skill-creator/scripts/security_review.py --path ~/.qclaw/skills/<your-skill-name>
AI Skill Creator Data Schema & Taxonomy
The skill organizes its development assets into a strict taxonomy to ensure clarity and security:
| File/Folder | Purpose | Requirement |
|---|---|---|
| SKILL.md | Primary instruction file for the AI agent | Must be < 500 lines |
| scripts/ | Executable Python, Bash, or Node.js logic | Must be tested and zero-error |
| references/ | Design docs, architecture, and security reports | For progressive disclosure |
| assets/ | Static resources and images | Isolated from core context |
| metadata | YAML frontmatter defining name, version, and OS | Validated during packaging |
AI Skill Creator Advanced Features
- Inter-Agent Task Interface: Supports cross-agent calling using isolated sessions to prevent cross-contamination.
- CVSS-Based Risk Assessment: Automatically calculates vulnerability scores to determine if a skill requires high-level CISO authorization.
- STRIDE Threat Modeling: Built-in modeling for Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of privilege.
- Automated Rollback Strategy: Integrated recovery procedures to revert to previous versions or Git tags if a quality gate or publishing step fails.
SKILL.md
Loading
Related Openclaw Skills
Wiki Query
A sophisticated tool for querying persistent markdown wikis and filing durable, synthesized results back into the knowledge base.
Karpathy Wiki
A framework for AI agents to build and maintain persistent markdown knowledge bases that bridge the gap between raw data and actionable intelligence.
Exec Tool
A security-focused command execution bridge for managing Openclaw Skills and system operations via controlled CLI triggers.
nano-img-cli
A specialized technical skill for managing the nano-img and nano-image CLI tools to drive Gemini-powered image generation and configuration directly from the terminal.
Wiki Ingest
A sophisticated workflow for ingesting new source material into persistent markdown wikis while maintaining structural integrity and entity clarity.
Wiki Lint
A markdown knowledge base auditor designed to identify structural inconsistencies, broken links, and content contradictions in persistent wikis.
Featured*
