How to Install Skills from Claw Mart in 2 Minutes

Most people overthink this.

They find Claw Mart, see a skill or persona they want, buy it, and then stare at their screen wondering what to do next. Where does the file go? How does OpenClaw know it exists? Is there some complicated setup process?

No. There isn't.

Installing a skill from Claw Mart into your OpenClaw agent takes about two minutes — and most of that is reading the file you just downloaded. I've installed dozens of skills across multiple agents at this point, and the process is dead simple once you see it once. Let me walk you through the entire thing so you never have to wonder.

What a "Skill" Actually Is

Before we get to the install process, let's clear up what you're actually buying. A Claw Mart skill isn't software in the traditional sense. There's no executable, no Docker container, no dependency hell. A skill is a structured markdown file — usually called SKILL.md — that gives your OpenClaw agent a new capability by teaching it a specific workflow, set of rules, or operational pattern.

Think of it like this: your OpenClaw agent already has general intelligence. It can reason, write code, send messages. What it lacks is your specific operational knowledge. A skill file fills that gap. It tells the agent exactly how to handle a particular domain — email security, coding sessions, memory management, Twitter posting, whatever — with battle-tested rules instead of vibes.

A persona is the same idea but bigger. Instead of one capability, it's an entire operating personality: a SOUL.md file that defines how the agent thinks, plus a bundle of skills that define what it can do.

That's it. Markdown files in the right folders. Let's get them there.

Step 1: Buy and Download from Claw Mart

Head to shopclawmart.com, find the skill you want, and buy it. After purchase, you'll get a download — usually a .zip or a set of .md files.

Unzip it if needed. You'll typically see something like this:

skill-download/
├── SKILL.md          # The main skill file
├── README.md         # Setup instructions (read this)
├── templates/        # Any supporting templates
│   └── MEMORY.md     # Example: memory template
└── config/           # Optional config files

Some skills are a single file. Some include supporting templates or scripts. The README will tell you what's what, but the core of every skill is that SKILL.md file.

Step 2: Find Your OpenClaw Workspace

Your OpenClaw instance has a workspace directory — this is where your agent reads its configuration, skills, and memory files. The typical structure looks something like this:

~/clawd/
├── SOUL.md              # Agent personality
├── AGENTS.md            # Access inventory & rules
├── MEMORY.md            # Agent memory
├── skills/              # Skill files live here
│   ├── HEARTBEAT.md
│   ├── EMAIL.md
│   └── CODING.md
├── notes/               # Daily notes
├── cron/                # Scheduled tasks
└── bin/                 # Scripts and CLIs

Your exact path might differ depending on your setup, but the concept is the same: there's a root directory, and skills go in the skills/ subdirectory (or sometimes at the root level, depending on the skill type).

If you're not sure where your workspace lives, check your OpenClaw configuration or just ask your agent: "Where is your workspace root?" It knows.

Step 3: Drop the Files In

Here's the "installation" process:

# Copy the skill file into your skills directory
cp ~/Downloads/skill-download/SKILL.md ~/clawd/skills/EMAIL_FORTRESS.md

# If there are templates, put them where they belong
cp ~/Downloads/skill-download/templates/MEMORY.md ~/clawd/MEMORY.md

That's it. That's the install.

No package manager. No build step. No restart required. Your OpenClaw agent reads its workspace files on every interaction, so the next time you talk to it, the skill is live.

For personas, you're usually replacing or merging your SOUL.md file:

# Back up your existing SOUL.md first
cp ~/clawd/SOUL.md ~/clawd/SOUL.md.backup

# Install the new persona
cp ~/Downloads/persona-download/SOUL.md ~/clawd/SOUL.md
cp ~/Downloads/persona-download/skills/* ~/clawd/skills/

Step 4: Configure (If Needed)

Some skills need light configuration. This is where the two minutes gets used. Let me give you some real examples from actual Claw Mart skills:

Example: Access Inventory

The Access Inventory skill ($5) comes with an AGENTS.md template. After dropping it in, you fill in your actual tools:

## Access Inventory

| Tool | Auth Method | Location | Status |
|------|------------|----------|--------|
| GitHub CLI | SSH key | ~/.ssh/id_ed25519 | ✅ Active |
| OpenRouter | API key | ~/.config/openrouter/key | ✅ Active |
| Vercel CLI | Token | ~/.vercel/token | ✅ Active |
| Railway CLI | Token | ~/.railway/token | ✅ Active |

This takes maybe 60 seconds. You list what your agent has access to, and it stops doing that infuriating thing where it claims it can't do something it absolutely can.

Example: X/Twitter Agent

The X/Twitter Agent skill ($9) needs API keys. You set up a keys file:

mkdir -p ~/.config/x-api
cat > ~/.config/x-api/keys.env << 'EOF'
X_API_KEY=your_api_key
X_API_SECRET=your_api_secret
X_ACCESS_TOKEN=your_access_token
X_ACCESS_TOKEN_SECRET=your_access_token_secret
X_USER_ID=your_user_id
EOF

Then drop the included xpost CLI script into your bin:

cp ~/Downloads/skill-download/xpost ~/clawd/bin/xpost
chmod +x ~/clawd/bin/xpost

Test it:

xpost post "Hello world"

Done. Your agent can now tweet. The skill file handles the content strategy, guardrails, scheduling cadence, and all the rules around what to post and what to never post.

Example: Coding Agent Loops

The Coding Agent Loops skill (free, by the way) needs nothing beyond the file drop. It teaches your agent how to run Codex and Claude Code in persistent tmux sessions with retry loops. No configuration — just new knowledge:

cp ~/Downloads/skill-download/SKILL.md ~/clawd/skills/CODING.md

Your agent now knows about Ralph loops, tmux session management, PRD-based workflows, and completion hooks. The next time you ask it to work on a coding task, it'll use these patterns automatically.

Step 5: Verify It's Working

After installing, give your agent a task that exercises the new skill. Don't be abstract about it — give it something concrete:

Memory skill installed? → "Save a note that our launch date is July 15th and store it as a project fact."
Email Fortress installed? → "Check my inbox and triage the last 10 emails."
Autonomy Ladder installed? → "What tier would you classify 'restarting a crashed background process' as?"
Coding Agent Loops installed? → "Set up a tmux session and run Codex against this PRD."

If the agent responds using the patterns from the skill (citing the right process, following the right steps), it's working. If it falls back to generic behavior, double-check your file is in the right directory and named correctly.

The Fast Path: Starter Packs and Personas

If you're standing up a new OpenClaw agent and want it to be useful immediately, installing skills one at a time is the slow way to do it. What you actually want is a bundle that handles the foundational stuff so you can start working, not configuring.

Felix's OpenClaw Starter Pack ($29) is the move here. It includes six pre-configured skills that cover the stuff every serious agent needs from day one:

Three-Tier Memory System — knowledge graph, daily notes, tacit knowledge. Your agent actually remembers things across sessions with structured layers instead of a flat text file that gets stale.
Coding Agent Loops — persistent tmux sessions with retry loops. Your agent can run coding tasks without crashing and losing work.
Email Fortress — email security that treats inbound messages as untrusted input. Nobody can prompt-inject your agent through your inbox.
Autonomy Ladder — the three-tier decision framework (act-and-report, act-and-detail, propose-and-wait) that solves the fundamental tension of wanting your agent to be independent without surprising you.
Access Inventory — so your agent always knows what tools it has.
Nightly Self-Improvement — your agent reviews its own performance at 3 AM and ships safe improvements while you sleep.

The install process is the same — download, unzip, copy files to your workspace — but you get a coherent system instead of piecemeal parts. All six skills are designed to work together. The memory system feeds the nightly improvement process. The autonomy ladder governs what the heartbeat monitoring can fix on its own. The access inventory means the coding loops actually know what tools are available.

It takes the same two minutes to install but saves you hours of figuring out which individual skills you need and how to make them play nice.

Handling Multiple Skills

One question people ask: what happens when you install a bunch of skills? Does the agent get confused?

No, because of how OpenClaw handles context. Each skill file is scoped to its domain. The EMAIL_FORTRESS.md file only activates when email is relevant. The CODING.md file only matters during development tasks. OpenClaw is smart about pulling in the right context at the right time — you don't need to worry about skills conflicting with each other.

That said, there's one thing to watch: if two skills give contradictory instructions about the same domain, the agent will try to reconcile them, and the results might be weird. This is why bundles like the Starter Pack are nice — the skills are pre-tested to work together. If you're mixing and matching individual skills from different authors, just skim them for conflicting rules before installing.

Setting Up Cron Jobs for Scheduled Skills

Some skills are designed to run on a schedule. The Nightly Self-Improvement skill ($9) runs at 3 AM. The Morning Briefing System ($5) triggers on your first check-in after 6 AM. The Business Heartbeat Monitor ($5) runs continuous health checks.

These skills include cron setup instructions in their README files. The typical pattern is:

# In your OpenClaw cron configuration
# Nightly self-improvement at 3 AM
0 3 * * * /path/to/openclaw run nightly-build

# Morning briefing on first interaction after 6 AM
# (handled via heartbeat pattern, not traditional cron)

# Heartbeat checks every 15 minutes
*/15 * * * * /path/to/openclaw run heartbeat

The skill files themselves define what happens during these runs. The cron just triggers when. Install the file, set up the cron, and your agent starts operating on autopilot.

The Full Stack: Going From Zero to Autonomous Agent

Here's what a complete setup looks like if you're going all-in. This is roughly the order I'd recommend:

Install the Starter Pack — gives you memory, coding, email security, autonomy framework, access tracking, and nightly improvement. Your agent goes from a blank slate to a functional operator.
Add the Morning Briefing System — now your agent starts every day by prepping your priorities. $5 and five seconds to install.
Add the Business Heartbeat Monitor — if you're running anything in production, your agent watches it 24/7. $5.
Add domain-specific skills — the SEO Content Engine ($29) if you're publishing content, the X/Twitter Agent ($9) if you want social media presence, Sentry Auto-Fix ($9) if you're shipping code.
Customize the SOUL.md — use the SOUL.md Design Kit ($5) to give your agent a real personality instead of default LLM blandness.

Or, if you want the everything-included option: the full Felix persona ($99) includes every skill in the catalog plus five months of production-hardened operational patterns. It's the most battle-tested configuration available on Claw Mart — built from running an actual business autonomously.

Common Mistakes

Putting files in the wrong directory. If your agent isn't picking up a skill, the file is probably not where OpenClaw expects it. Check your workspace structure.

Not reading the README. Seriously. It's one page. It tells you exactly what goes where and what to configure.

Installing too many skills before testing any of them. Drop one in, test it, confirm it works, then add the next. You'll debug much faster.

Overriding SOUL.md without backing up. If you've customized your agent's personality and then install a persona that replaces SOUL.md, your customizations are gone. Back it up first.

What You Should Do Right Now

If you've been thinking about making your OpenClaw agent more capable but haven't pulled the trigger, just start. Pick one skill — the Coding Agent Loops is free, so there's literally no excuse — download it, copy it into your workspace, and test it.

If you want the full foundation in one shot, grab the Felix's OpenClaw Starter Pack. Twenty-nine bucks, six skills, two minutes to install, and your agent immediately becomes meaningfully more capable.

The whole point of Claw Mart is that you shouldn't have to figure out every operational pattern from scratch. Someone already hit the edge cases, already got prompt-injected through email, already lost work to crashed coding sessions, already dealt with agents that forget everything between conversations. The skills encode those lessons so you don't have to learn them the hard way.

Two minutes. That's all it takes.