clawhub-skill-creator（ClawHub 技能 creator）

Clawhub Skill Creator

Complete guide for creating skills compatible with clawhub registry and optimized for AI agent usage.

Language Rule

Skill description must be in English, unless context requires otherwise (examples, comments, etc. may use other languages when appropriate).

• description in frontmatter: English only
• SKILL.md body: English preferred
• Examples, code comments: Any language as needed
• References: Any language as appropriate for domain

Quick Start

Create New Skill

# Run initializer
./scripts/init-skill.sh my-skill

# Or manually create structure
mkdir -p my-skill/{references,scripts,assets}
touch my-skill/SKILL.md my-skill/_meta.json my-skill/LICENSE.txt

Skill Creation Workflow

Phase 1: Understand Requirements

Before writing any code, clarify:

1. What problem does this skill solve?

- Concrete use cases (2-3 examples) - Target queries that should trigger skill

1. Who will use this skill?

- AI agents (primary audience for clawhub) - Expected context (tools available, environment)

1. What resources are needed?

- References for detailed docs? - Assets for templates? - Scripts for automation?

Output: Clear understanding of skill scope and triggers.

Phase 2: Plan Structure

Choose pattern based on complexity:

Decide on resources:

• Which references needed? (docs, examples, patterns)
• Which assets needed? (templates, configs)
• Which scripts needed? (validation, packaging)

Output: Directory structure and resource list.

Phase 3: Initialize Structure

# Create directory
mkdir -p my-skill/{references,scripts,assets}

# Create required files
touch my-skill/SKILL.md
touch my-skill/_meta.json
touch my-skill/LICENSE.txt

Required files for clawhub:

my-skill/
├── SKILL.md              # Instructions and metadata (required)
├── _meta.json            # Registry metadata (required)
├── LICENSE.txt           # License file (required)
├── references/           # Optional: detailed documentation
├── scripts/              # Optional: automation scripts
└── assets/               # Optional: templates, resources

Phase 4: Write SKILL.md

Frontmatter (YAML)

---
name: skill-name
description: What it does. Use when: (1) trigger-1, (2) trigger-2, (3) trigger-3.
---

Critical rules:

• name must match directory name
• description is ONLY trigger mechanism — include all "when to use" here
• description must be in English
• Only name and description are required in frontmatter

Body Structure

# Skill Title

Brief purpose (1-2 sentences).

## Quick Start

**Linux/Mac:**

command --option

**Windows CMD:**

command --option

**PowerShell:**

command --option

## When to Use

- Situation 1: What to do
- Situation 2: What to do
- Situation 3: What to do

## Workflow

1. **Step one**: Description
2. **Step two**: Description
3. **Step three**: Description

## Resources

- `references/advanced.md` - For complex cases
- `references/examples.md` - Usage examples
- `assets/template.txt` - Starting template

Writing guidelines:

• Imperative voice ("Open file", not "You should open file")
• Concrete examples over abstract explanations
• Platform-aware commands
• Navigation to references
• English preferred for body text

Phase 5: Create _meta.json

{
  "name": "skill-name",
  "version": "1.0.0",
  "description": "Short description for registry listing",
  "requires": {
    "env": ["ENV_VAR_1", "ENV_VAR_2"],
    "credentials": ["credential_name"]
  },
  "tags": ["tag1", "tag2", "latest"]
}

Fields explained:

• name: Must match directory and SKILL.md frontmatter
• version: Semver (X.Y.Z), check registry before setting
• description: For registry listing, must be in English
• requires.env: Environment variables needed
• requires.credentials: Credentials needed
• tags: Include "latest" for discoverability

Phase 6: Add LICENSE.txt

Choose license (MIT recommended):

MIT License

Copyright (c) 2025 [Author]

Permission is hereby granted...

Phase 7: Write References (if needed)

Create files in references/:

# Reference Title

Detailed documentation here.

## Section

Content...

Guidelines:

• One topic per file
• <5K words per file
• Link from SKILL.md with clear context
• No deeply nested references
• Language: English preferred, domain-specific allowed

Phase 8: Validate Locally

# Check structure
./scripts/validate.sh my-skill

# Or manual checks:
# - SKILL.md exists and has frontmatter
# - _meta.json is valid JSON
# - LICENSE.txt exists
# - No README.md, CHANGELOG.md
# - Line count < 300
# - References linked correctly

Validation checklist:

• [ ] Directory name matches name in frontmatter and _meta.json
• [ ] description in English, includes "Use when:" triggers
• [ ] SKILL.md < 300 lines
• [ ] _meta.json valid JSON
• [ ] No extraneous files (README, CHANGELOG)
• [ ] References exist and linked
• [ ] Token estimate < 10K

Phase 9: Test with Agent

Trigger test:

• Does skill activate for intended queries?
• Does description correctly trigger skill?

Workflow test:

• Can agent follow steps without clarification?
• Are commands clear and executable?

Resource test:

• Are references loaded at appropriate time?
• Is navigation clear?

Edge case test:

• How does skill handle errors?
• Are platform differences handled?

Phase 10: Iterate (if needed)

If tests reveal issues:

1. Identify problem

- Trigger not working? → Fix description - Workflow unclear? → Rewrite steps - Missing info? → Add reference

1. Update files

- SKILL.md, _meta.json, or references

1. Re-validate and re-test

- Go back to Phase 8

1. Repeat until satisfied

Iterate cycle: Phase 8 → Phase 9 → Phase 10 (loop) → Phase 11

Phase 11: Check Version

Before publishing, verify current registry version:

# Check current registry version
clawhub inspect skill-name --json | grep version

# Ensure new version follows semver:
# 1.0.0 → 1.0.1 (patch: bug fixes)
# 1.0.0 → 1.1.0 (minor: new features)
# 1.0.0 → 2.0.0 (major: breaking changes)

# Never downgrade! (1.1.0 → 1.0.2 is wrong)

Update _meta.json with correct version:

{
  "version": "1.0.1"
}

Phase 12: Package

# Create .skill package for distribution
./scripts/package-skill.sh my-skill ./dist

# Output: dist/my-skill.skill
# Validates structure before packaging

Package contains:

• All skill files
• Validated structure
• Ready for distribution

Phase 13: Publish

cd my-skill

# Publish to clawhub
clawhub publish . --version 1.0.1 --changelog "Description of changes"

# Verify publication
clawhub inspect skill-name

After publish:

• Skill available in registry
• Others can install via clawhub install skill-name

Required Files for Clawhub

SKILL.md

Instructions and metadata:

• YAML frontmatter (name, description in English)
• Markdown body (workflow, examples)
• Navigation to references

_meta.json

Registry metadata:

{
  "name": "skill-name",
  "version": "1.0.0",
  "description": "Registry listing description in English",
  "requires": {
    "env": [],
    "credentials": []
  },
  "tags": ["latest"]
}

LICENSE.txt

License file (MIT, Apache-2.0, etc.)

Resources

• references/skill-structure.md - Directory structure patterns
• references/agent-first-design.md - Designing for AI vs humans
• references/token-optimization.md - Minimizing context usage
• references/cross-platform.md - Platform-aware scripts
• references/validation-checklist.md - Pre-publish checks
• references/versioning.md - Semver best practices

Key Principles

1. Agent-First Design

Skills used by AI agents, not humans:

• ❌ No interactive prompts
• ❌ No platform-specific scripts (use knowledge instead)
• ✅ Command templates for all platforms
• ✅ Clear navigation to references

2. Progressive Disclosure

Level 1: Metadata (name + description)     → Always loaded
Level 2: SKILL.md body                      → On trigger
Level 3: Resources (references/, assets/)   → On demand

3. Token Budget

4. Cross-Platform Awareness

Instead of script:

## Commands

**Linux/Mac:**

command --option

**Windows CMD:**

command --option

**PowerShell:**

command --option

Agent chooses appropriate variant based on detected platform.

5. English Language for Descriptions

Required in English:

• description in SKILL.md frontmatter
• description in _meta.json
• Main workflow instructions

May use other languages:

• Code examples
• Comments
• Domain-specific references
• User-facing examples

Scripts

• scripts/init-skill.sh - Initialize new skill structure
• scripts/package-skill.sh - Package skill for distribution
• scripts/validate.sh - Validate skill structure

Anti-Patterns

❌ Don't:

• Put "When to use" only in body (must be in description)
• Use non-English description in frontmatter
• Duplicate info between SKILL.md and references
• Create README.md, CHANGELOG.md (clutter)
• Use platform-specific scripts (sh/bat)
• Write passive voice ("You should")
• Include generic background theory
• Skip validation before publish
• Forget to bump version
• Publish without testing

✅ Do:

• Write description in English
• Start with concrete examples
• Move details to references/
• Use imperative voice ("Do X")
• Challenge every sentence's value
• Test with real agent queries
• Validate before publish
• Follow semver strictly
• Iterate based on test results