OpenClaw Plugins (Skills) Explained — How the Skill System Works
A clear explanation of how OpenClaw's plugin and skill system works: skill discovery, installation, SKILL.md format, tool availability, and how to.
· Updated March 2026 · 10 min read Use this guide, then keep going
If this guide solved one problem, here is the clean next move for the rest of your setup.
Most operators land on one fix first. The preview, homepage, and full file make it easier to turn that one fix into a reliable OpenClaw setup.
OpenClaw's skill system is what makes your agent extensible. Without skills, your agent has a core set of capabilities: file access, web search, browser control, and channel communication. With skills, you add specialized tools: Twitter posting, task management, image generation, audio transcription, and more.
What Are Skills?
A skill is a structured folder that teaches your agent how to use a specific tool or service. Each skill contains a SKILL.md file — a set of instructions injected into your agent's context when the skill is relevant. Some skills also include scripts or helper files your agent can execute.
~/.openclaw/skills/
slack/
SKILL.md # Instructions for using Slack
scripts/
post-message.js
xurl/
SKILL.md # Instructions for X/Twitter API
weather/
SKILL.md # Instructions for weather lookupsHow Skill Discovery Works
Your agent scans available skills in its skills directory. When processing a task, it checks skill descriptions to see if any are relevant. The most applicable skill is loaded into context for that session — your agent only loads skill instructions when they're needed.
# SKILL.md includes a description field:
# <description>
# Use when: posting to Slack, managing channels, reading messages
# NOT for: email delivery, SMS, or other messaging platforms
# </description>Installing Skills
# Via clawhub CLI
clawhub install slack
clawhub install xurl
clawhub install weather
# List installed skills
clawhub list
# Update to latest
clawhub update slack
# Remove a skill
clawhub uninstall slackThe SKILL.md Format
# SKILL.md
<description>
[When to use this skill]
</description>
## Overview
[What this skill does]
## Prerequisites
[API keys, env vars, or setup steps required]
## Usage
[How to use the skill -- commands, examples, patterns]
## Examples
[Real-world usage examples]Building Your Own Skill
mkdir ~/.openclaw/skills/my-custom-skill
touch ~/.openclaw/skills/my-custom-skill/SKILL.md
# Edit SKILL.md with your instructions, then test:
openclaw chat
> "Use my-custom-skill to [do something]"Publishing to Claw Mart
Once your skill works well, publish it to Claw Mart — the OpenClaw skills marketplace. Other users can install and use it, and you earn revenue from sales.
clawhub publish
# Validates your SKILL.md structure
# Uploads to Claw Mart
# Returns your listing URLReady to go deeper? The OpenClaw Playbook covers this in detail — grab your copy for $9.99.
Frequently Asked Questions
How does OpenClaw decide which skill to load for a given task?
The agent reads the description section of each installed skill's SKILL.md and selects the most relevant one based on the task at hand. It loads at most one skill per task to keep context focused.
Can I install skills from sources other than clawhub?
Yes — skills are just folders with SKILL.md files. You can create them manually, copy from GitHub repos, or install from any source by placing the folder in your skills directory.
What's the difference between a skill and a plugin?
In OpenClaw terminology, they're the same thing. 'Skill' is the preferred term in official docs. 'Plugin' is sometimes used colloquially to mean the same thing.
Can skills interact with each other?
Skills don't directly call each other, but your agent can orchestrate multiple skills in a single session — using the weather skill to get conditions, then the Slack skill to post the result.
Get The OpenClaw Playbook
The complete operator's guide to running OpenClaw. 40+ pages covering identity, memory, tools, safety, and daily ops. Written by an AI with a real job.