Understanding AgentSkills

Core Concept

AgentSkills are modular, self-contained packages that extend OpenClaw's capabilities by providing specialized knowledge, workflows, tools, and assets. They transform OpenClaw from a general-purpose agent into a specialized agent with procedural knowledge.

Anatomy of a Skill

Every skill has a standard structure:

skill-name/
├── SKILL.md          # Required (metadata + instructions)
├── scripts/          # Optional: executable code
├── references/       # Optional: detailed documentation
└── assets/           # Optional: templates, static files

SKILL.md (Required)

Contains YAML frontmatter with name and description (used for skill discovery), followed by Markdown instructions.

Frontmatter Rules:

• name: 1-64 chars, lowercase alphanumeric-hyphens only
• description: 1-1024 chars, must include "Use when..." for context

Scripts (Optional)

Executable code (Python/Bash) for tasks that need deterministic reliability or are repeated frequently.

References (Optional)

Extended documentation loaded on-demand when needed.

Assets (Optional)

Templates, images, or other static files used in output.

Creating a New Skill

1. Create directory: mkdir skill-name
2. Create SKILL.md with required frontmatter:

---
name: skill-name
description: Brief description. Use when doing X.
---
# Skill Title

## Instructions