Skills and slash commands solve two different problems. Skills solve "my agent can't do X" — they add integrations, tools, and behaviors that aren't built in. Slash commands solve "I keep typing the same long instruction" — they give you a two-word shortcut to any behavior your agent knows.
Most builders learn slash commands first (because they're immediately useful), then discover the skill system when they need something that doesn't exist yet. Both are worth mastering.
Skills vs Slash Commands: The Distinction That Matters
A skill is a module that extends your agent's capabilities. It might add tools (web search, file handling), integrations (Gmail, Notion, Jira), or behaviors (a specialized reasoning mode). Skills are installed via the CLI and run as part of the agent's execution context.
A slash command is a shortcut trigger. When a user types /reset, the agent interprets it as an instruction to clear context. Slash commands don't add capabilities — they provide fast access to capabilities that already exist.
The overlap: skills can register their own slash commands. When you install the Gmail skill, it registers /gmail-check and /gmail-send. The commands come with the skill but are conceptually separate from it.
Installing and Managing Skills
# Install from ClaWHub registry
openclaw skills install web-search
# Install from local directory
openclaw skills install --path ./my-custom-skill
# List installed skills
openclaw skills list
# Remove a skill
openclaw skills remove web-searchAfter installing a skill, restart the gateway. Skills that modify the agent's tool set require a gateway restart to take effect. Skills that only add slash commands may work without a restart in newer gateway versions — but restarting is always safe.
Built-In Slash Commands Reference
These commands work in every setup without installing any skills:
| Command | What It Does |
|---|---|
/help |
Lists all available slash commands with descriptions |
/reset |
Clears the current conversation context |
/memory |
Shows current agent memory contents |
/status |
Reports agent and channel health |
/model |
Switches the active LLM for this session |
/version |
Returns current OpenClaw version |
Creating Custom Slash Commands Without Code
The fastest way to create custom slash commands is in soul.md. Under the commands section, define a name and what the agent should do when that command is triggered:
## Commands
/brief
Generate a concise morning briefing covering: pending tasks,
messages since yesterday, and today's priorities. Format as
bullet points. Keep it under 200 words.
/standup
Summarize what I've worked on in the past 24 hours based on
our conversation history. Format for a team standup: yesterday,
today, blockers.
/focus
Enter focus mode: respond only to task-related queries, skip
social chat, and remind me of my current priority if I go off-track.These commands are pure natural language instructions. No coding, no config files beyond soul.md. The agent interprets them as behavioral prompts when triggered.
Skill-Registered Slash Commands
Skills that register slash commands document them in their README or ClaWHub listing. Here's a typical pattern from the web-search skill:
/search [query] — Run a web search and return top results
/search-news [topic] — Search recent news articles only
/search-academic [term] — Search academic sources onlySkill commands accept arguments after the command name. The skill's handler parses the argument and uses it to parameterize the action. Not all commands require arguments — some operate on conversation context without any explicit input.
To see all commands currently registered (built-in + skills + custom), type /help in any connected channel. The output dynamically reflects your current skill installation.
Common Mistakes
Installing a skill but not restarting the gateway
The skill is installed but not active. Its slash commands won't appear in /help and its tools won't be available until the gateway restarts. This catches people out every time.
Defining custom commands with vague instructions
A custom command like /report — generate a report is too vague to produce consistent output. Write the command instruction as if you're giving the agent a detailed task brief. More specificity produces better results.
Using the same command name across multiple contexts
If you install two skills that both register /search, one wins and the other is silently ignored. Choose skills with non-overlapping command namespaces, or use custom commands in soul.md to override with your preferred behavior.
Frequently Asked Questions
What is the difference between a skill and a slash command in OpenClaw?
A skill is a module that adds capabilities to your agent — tools, integrations, or behaviors. A slash command is a shortcut that triggers specific agent behavior when typed as /commandname in a chat. Skills can register slash commands, but not all slash commands come from skills. Built-in slash commands exist independently of the skill system.
How do I install a skill in OpenClaw?
Run openclaw skills install [skill-name] to install from ClaWHub, or openclaw skills install --path ./my-skill to install from a local directory. After installation, restart the gateway for the skill to activate. Verify with openclaw skills list.
How do I see all available slash commands?
Type /help in any connected channel to get a list of all registered slash commands with brief descriptions. Alternatively, run openclaw skills list to see installed skills — each skill's documentation lists its slash commands. Built-in commands always appear regardless of installed skills.
Can I create my own slash commands?
Yes. Define custom slash commands in your soul.md file under the commands section, or create a skill that registers commands programmatically. Custom commands in soul.md are the fastest way to add shortcuts — no coding required. They execute as agent instructions when triggered.
What are the most useful built-in slash commands?
/reset clears conversation context, /memory shows current memory contents, /status checks agent and channel health, /help lists all commands, and /model switches the active LLM without restarting. These built-ins work in every channel without any skill installation.
Do slash commands work in all channels?
Built-in slash commands work in all channels that support text input. Some channels (like voice or SMS) may not support slash command syntax — the agent treats the text literally. Skill-registered commands follow the same rules. Check your channel's documentation for slash command support status.