Skill Starter

Generate a reviewable starting point for a new OpenClaw skill.

When to use

• You want a repeatable folder layout instead of assembling a skill by hand.
• You need a starter SKILL.md, smoke test, fixtures, and optional CI workflow.
• You want a stricter template with Safe Install policy scaffolding.

Command

node {baseDir}/bin/create-openclaw-skill.js my-skill
node {baseDir}/bin/create-openclaw-skill.js my-skill --no-prompts --out ./skills
node {baseDir}/bin/create-openclaw-skill.js my-skill --template strict-security --ci
node {baseDir}/bin/create-openclaw-skill.js my-skill --force

Templates

Generated files

• SKILL.md: skill instructions with OpenClaw-compatible frontmatter
• package.json: project manifest
• docs/: architecture and security notes
• scripts/: local script directory
• tests/: smoke test stub
• .env.example: environment variable template
• CHANGELOG.md: release notes starter
• scripts/profile-target.js: profiling helper
• fixtures/: sample input data
• .github/workflows/security-scan.yml: optional CI workflow

Interactive prompts

strict-security extras

• .openclaw-tools/safe-install.policy.json: Safe Install policy scaffold
• docs/security.md: security integration guide
• .github/workflows/security-scan.yml: ClawShield scan workflow

Boundaries

• The scaffold is intentionally lightweight. You still need to replace placeholders and add real business logic.
• Generated SKILL.md files are now aligned to OpenClaw-style frontmatter and use {baseDir} for local command examples.