skill-creater

Generate complete AgentSkills from user requirements. Creates SKILL.md, scripts, references, assets folders, and packages them into a ready-to-upload archive...

Security Scan

The skill's purpose (generate and package skills) mostly matches the included code, but there are multiple inconsistencies and overstated behaviors that make the package look sloppy or incomplete — not clearly malicious, but you should inspect and test before trusting it.

Purpose & Capability

The skill claims to 'interview' users, generate complete SKILL.md, scripts, references, and package into a zip. The repository contains generator and packager scripts that implement name/manifest generation and zipping, which aligns with the stated purpose. However the claimed interactive interview and some 'automatic assurances' are not implemented in the provided scripts; overall capability is plausible but overstated.

Instruction Scope

SKILL.md describes an automated interview flow and many guarantees (YAML compliance, progressive disclosure). The code (init_skill_template.py) is a simple CLI that accepts a single description argument — it does not implement an interactive interview or advanced validation. The packager validates only basic frontmatter existence and a few manifest fields. The instructions therefore promise more automation and validation than the code provides.

Install Mechanism

No install spec; this is an instruction-and-code skill only. There are no external downloads, no network installs, and no injected third-party packages referenced. Packaging and file I/O are local.

Credentials

The skill requests no environment variables, no credentials, and references no external secrets. The code only reads/writes local filesystem paths and manifest/SKILL.md files, which is consistent with the purpose.

Persistence & Privilege

The skill does not request persistent presence (always: false) and does not modify other skills or system-wide settings. It runs local file operations only and is not configured for forced inclusion.

What to consider before installing

This package appears to be a developer utility but has several red flags that point to sloppy or incomplete implementation rather than clear malicious intent. Things to do before trusting or using it: - Inspect and run the Python scripts locally in a safe environment (sandbox/VM) and review output ZIP contents before uploading anywhere. - Note naming/path inconsistencies: SKILL.md and manifest mention 'skill-creater' but build.sh and some human-facing text say 'skill-craftsmen'/'skill-craftsmen' and build.sh calls scripts/package_skill.py while package_skill.py is at repository root. This likely indicates broken scripts or copy-paste errors you should fix before use. - Expect the 'interview' capability to be manual: init_skill_template.py currently only accepts a single CLI argument (no interactive Q&A). If you need interactive prompting, add or review that logic yourself. - Validate generated SKILL.md and manifest.json contents before uploading to Clawhub; the packager's validation is minimal. - Because the package writes files and zips directories, do not run it on directories containing sensitive files (or run it with a dedicated temp folder) to avoid accidental inclusion. - If you plan to let an autonomous agent invoke this skill, consider restricting autonomous invocation until you've verified the generated outputs are safe and correct. If you want, I can: (1) point out exact lines to change to fix build.sh path/name mismatches, (2) show a short test plan to validate the packager locally, or (3) produce a hardened checklist for reviewing generated skills before upload.

Skill Craftsmen

A meta-skill that generates other skills. Transform your ideas into production-ready AgentSkills with proper structure, documentation, and packaging.

What This Skill Does

1. Interviews you to understand what skill you want to build
2. Generates complete skill structure including SKILL.md, scripts, references
3. Packages everything into a ready-to-upload .zip archive
4. Provides Clawhub upload guidance

Quick Start

User: "帮我创建一个处理PDF的Skill"
→ This skill interviews → Generates → Packages → Delivers zip file

Usage Flow

Step 1: Intent Capture (自动)

Skill会自动询问关键问题：

• 这个skill要解决什么问题？
• 用户会怎么描述需求来触发它？
• 需要哪些功能？简单的还是复杂的？
• 需要脚本/工具吗？还是纯文本指导？

Step 2: Structure Generation

根据需求生成：

your-skill/
├── SKILL.md              # 主文档 (name, description, instructions)
├── manifest.json         # Clawhub元数据
├── scripts/              # 可选: 可执行脚本
│   └── (Python/Node.js/Bash)
├── references/           # 可选: 参考文档、示例
│   └── (schemas, examples, guides)
├── assets/               # 可选: 模板、图标等
│   └── (templates, icons, images)
└── README.md             # 安装说明 (可选)

Step 3: Package & Deliver

自动打包为 your-skill-v1.0.0.zip，包含：

• 完整目录结构
• 验证过的manifest.json
• 可直接上传Clawhub

Skill Complexity Levels

级别	特点	包含内容
Simple	纯文本指导	SKILL.md only
Standard	带辅助脚本	SKILL.md + scripts/
Advanced	完整工具集	SKILL.md + scripts/ + references/ + assets/

Best Practices Applied

This skill automatically ensures:

• ✅ YAML frontmatter format compliance
• ✅ Description includes "when to use" triggers
• ✅ Proper directory structure
• ✅ Concise SKILL.md (<500 lines)
• ✅ Progressive disclosure design

Output Example

📦 Generated: pdf-master-v1.0.0.zip

📁 Contents:
├── SKILL.md              ✓ Valid YAML frontmatter
├── manifest.json         ✓ Clawhub ready
├── scripts/
│   ├── merge.py
│   ├── split.py
│   └── rotate.py
├── references/
│   ├── examples.md
│   └── api-guide.md
└── README.md

🚀 Next Steps:
1. Extract the zip
2. Test locally: openclaw skill install ./pdf-master
3. Upload to Clawhub: https://clawhub.com/upload

Clawhub Upload Checklist

Generated skills include:

• Valid manifest.json with version, author, tags
• SKILL.md with proper YAML frontmatter
• Scripts are executable (if included)
• No unnecessary files (README optional)
• Compressed with proper structure

---

Craft your skills. Share with the world.