Best OpenClaw Skills in 2026: 10 Essential Skills From ClawHub Worth Installing

This updated guide reframes Best OpenClaw Skills in 2026: 10 Essential Skills From ClawHub Worth Installing around practical search intent: what readers need to compare, choose, install, secure, or operationalize in 2026. It focuses on decision criteria, workflow fit, and the trade-offs that matter once an AI agent, skill, marketplace, or automation moves from curiosity to daily use.

The article also broadens the semantic coverage around SKILL.md, AI agent skills, agent instructions. That gives readers a clearer path from high-level research to implementation planning, while keeping the content useful for teams evaluating AI agent skill design.

Quick Answer

A useful skill is narrow, repeatable, and explicit about inputs, tools, constraints, and success criteria, so the agent can act consistently instead of guessing.

ClawHub surpassed 13,000 skills in February 2026. That sounds impressive until you discover that a Snyk security audit flagged 13.4% of them for critical issues — malware, prompt injection, exposed API keys. A separate Koi Security scan of 2,857 skills identified 341 actively stealing user data.

So the real question is not "what skills are available?" It is "which ones are safe and genuinely useful?"

We installed and tested over 40 skills from ClawHub across the past two weeks. Most did not survive day one — broken installs, silent permission grabs, or features that simply failed to work as advertised. Here are the 10 that held up — each one addresses a specific problem, has a maintained codebase, and will not leak your credentials. Install any of them with:

npx clawhub@latest install <skill-name>

1. Web Browsing — The Foundation

Install: npx clawhub@latest install web-browsing

Without web access, your OpenClaw agent is essentially a chatbot operating on stale training data. The official web browsing skill gives it the ability to navigate pages, extract content, and follow links — the foundation that most other workflows depend on.

It is the most-installed skill on ClawHub (180,000+ installs) for good reason. Research tasks, documentation lookups, fact-checking — they all begin here.

One limitation: it fetches pages but does not search. For that, you need a dedicated search skill.

Good for: Research, documentation, fact-checking, content extraction.

2. Felo Search — Real-Time AI Search With Source Citations

Install: npx clawhub@latest install felo-search

Several search skills compete on ClawHub — Tavily, Exa, web-search-plus. Felo Search takes a distinct approach: rather than returning a list of links for your agent to parse, it delivers AI-synthesized answers with source citations attached.

Ask "what changed in React 19.1?" and you receive a direct answer with references, not 10 blue links to sift through. In our testing, this reduced the back-and-forth between search and action from 3-4 agent turns down to one.

What stands out:

• Delivers structured answers, not raw link dumps
• Cites sources for every claim — your agent can verify before acting
• Handles Chinese, Japanese, Korean, and English queries natively
• Free during the current open access period

The multi-language support is a genuine differentiator. Most search skills only perform well in English. If you work across languages, this eliminates the need for maintaining separate search tools per locale.

Setup: Requires a FELO_API_KEY — grab one free at felo.ai (Settings -> API Keys).

Good for: Developer docs, news monitoring, competitive research, multi-language queries.

3. Telegram Integration — Your Agent on Mobile

Install: npx clawhub@latest install telegram

The second most-installed skill on ClawHub (145,000+ installs). Connect your OpenClaw to Telegram with a BotFather token, and you can message your agent from your phone.

The practical value is straightforward: you are away from your desk, need to check something, and your agent handles it through Telegram. Setup takes roughly 5 minutes. Latency is surprisingly low for most queries.

Good for: Mobile access, quick queries on the go, team notifications.

4. Felo SuperAgent — Streaming Conversations With Built-In Tools

Install: npx clawhub@latest install felo-superAgent

This skill does not appear in other "best skills" lists because it launched recently — and there is nothing quite like it on ClawHub.

Most OpenClaw skills handle one thing: search the web, send a message, query a database. Felo SuperAgent is a multi-tool creative workstation packed into a single skill. It connects to Felo's SuperAgent API and gives your agent:

Real-time SSE streaming — responses appear as they generate, token by token. LiveDoc canvas — every conversation ties to a persistent workspace at felo.ai/livedoc. Generated images, documents, and slides are saved there automatically. 6 built-in tools — image generation, deep research reports, document creation, PPT generation, HTML pages, and X/Twitter search. 3 specialized modes — tweet writing (twitter-writer), logo design (logo-and-branding), and e-commerce product images (ecommerce-product-image). Multi-turn context — conversations carry state across messages, so you can iterate.

The LiveDoc integration is what distinguishes this from "just another chat skill." When your agent generates a research report or a slide deck mid-conversation, it does not disappear into the terminal — it is saved to a persistent canvas you can open in a browser, edit, and share.

A few things to know upfront: SuperAgent requires the felo-livedoc skill as a dependency (for LiveDoc management), and you need the same FELO_API_KEY used by other Felo skills. Image and design generation tasks benefit from a longer timeout setting (90 seconds vs the default 60).

# Install both SuperAgent and its dependency

npx clawhub@latest install felo-superAgent

npx clawhub@latest install felo-livedoc

Good for: Creative workflows, content generation, research sessions, multi-step projects where you want your agent to produce actual assets — not just text.

5. n8n Workflow — Connect Everything

Install: npx clawhub@latest install n8n-workflow

If you already use n8n for business automation, this skill transforms your OpenClaw agent into the natural language interface for your existing workflows. Trigger automations, pass data between steps, orchestrate multi-tool processes — all through conversation.

The value scales with how much you have already built in n8n. For someone with 50 workflows, this skill is transformative. For someone starting from scratch, the setup overhead might not justify itself immediately.

Good for: Business process automation, multi-step workflows, connecting APIs without code.

6. Felo Slides — AI Presentation Generation

Install: npx clawhub@latest install felo-slides

Here is a gap in the OpenClaw ecosystem that surprised us during testing: across 30+ skill recommendation articles and community discussions, not a single one mentions a presentation skill. PPT generation is a complete blind spot on ClawHub.

Felo Slides fills this gap. Describe what you want, and it generates a professional PPTX through Felo's PPT API. The skill manages the async process — submits the task, polls for completion, returns the download link and a LiveDoc URL for browser-based editing.

The generated slides use HTML-based rendering, meaning each element (text, images, shapes) is independently editable — unlike image-based slide generators where each slide is a flat picture.

npx clawhub@latest install felo-slides

"Create a 10-slide deck about our Q1 results"

-> Returns PPTX download link + LiveDoc editing URL

Good for: Meeting prep, pitch decks, training materials, any situation where you need slides and do not want to open PowerPoint.

7. GitHub Integration — Code Workflow in Natural Language

Install: npx clawhub@latest install github

Manage repos, issues, PRs, and code reviews through your agent. "Create an issue for the login bug" or "show me open PRs on main" maps to GitHub API calls without you touching the browser.

Not flashy, but it eliminates the constant context-switching between your agent and the GitHub UI. Particularly useful if you manage multiple repositories.

Good for: Issue tracking, PR management, code review workflows.

8. Database Query — Talk to Your Data

Install: npx clawhub@latest install database-query

Connect your agent to PostgreSQL, MySQL, or SQLite and query in natural language. "How many users signed up last week?" becomes SQL, gets executed, and comes back formatted.

A necessary warning: this skill runs queries against your actual database. Use a read-only connection string. Restrict to non-production databases unless you fully understand the permission model. The 95,000+ installs suggest plenty of people find it useful — just exercise caution with credentials.

Good for: Data analysis, quick reporting, ad-hoc queries without writing SQL.

9. Eleven Labs Agent — Voice for Your AI

Install: npx clawhub@latest install elevenlabs-agent

Text-to-speech using Eleven Labs' API. Your agent can read responses aloud, generate audio content, or create voiceovers. Requires a separate Eleven Labs API key.

Niche, but if you are building voice-enabled workflows or need accessibility features, it is the cleanest TTS integration on ClawHub.

Good for: Audio content, accessibility, voice-enabled workflows.

10. Home Assistant — Smart Home Control

Install: npx clawhub@latest install home-assistant

For Home Assistant users: control devices, check sensors, trigger automations through natural language. "Turn off the living room lights" or "what's the bedroom temperature?" — your agent becomes a keyboard-free smart home controller.

Only relevant if you run Home Assistant, but for that audience, it is a natural fit.

Good for: Smart home automation, IoT device management.

How to Install Skills Safely

The security situation on ClawHub is real. Before installing anything:

Check the source code. Every skill is a folder with a SKILL.md file. Read it. Look at what permissions it requests and what shell commands it runs. A "weather skill" that wants Bash(*) access is a red flag.

Verify the publisher. Skills from verified organizations (OpenClaw official, established companies) carry less risk than anonymous uploads. Check the GitHub repo — is it maintained? Does it have issues and PRs?

Watch for these warning signs:

• No GitHub repository linked
• Requests wildcard shell permissions
• Last updated more than 3 months ago
• Fewer than 100 installs with no reviews
• Skill name mimics a popular skill (typosquatting)

Keep skills updated. The ecosystem moves fast. Security patches matter.

# Core commands

npx clawhub@latest install <skill-name> # Install

npx clawhub@latest list # List installed

npx clawhub@latest update # Update all

npx clawhub@latest remove <skill-name> # Remove

The Felo Skills Ecosystem

Three skills on this list — Felo Search (#2), Felo SuperAgent (#4), and Felo Slides (#6) — come from the same provider. They share a single API key and function as an integrated toolkit.

The full family includes 8 skills:

All free, all MIT-licensed, all installable from ClawHub or via npm install -g felo-ai.

The integration between skills is where the value compounds. A search result from felo-search can feed into felo-slides to generate a presentation. Or felo-web-fetch grabs a competitor's page and felo-content-to-slides transforms it into a deck. These workflows happen within the same agent session using one API key.

Get started: felo.ai -> Settings -> API Keys

What Makes a Good OpenClaw Skill?

After testing dozens of skills from ClawHub, the pattern is consistent:

Focused scope. The strongest skills do one thing well. "AI everything" skills are usually "AI nothing" — bloated, slow, and poorly maintained.

Graceful error handling. Your agent should not crash because a skill returned unexpected data. Quality skills fail with clear messages, not silent breakage.

Minimal permissions. No unnecessary shell access, no phoning home, no storing credentials in plaintext. If a skill asks for more than it needs, skip it.

Active maintenance. The OpenClaw ecosystem evolves quickly. A skill that worked in January might break by March. Check the last commit date before installing.

Real documentation. If the SKILL.md is three lines long, the skill probably is not production-ready. Look for usage examples, parameter descriptions, and error handling notes.

Getting Started

If you are setting up OpenClaw for the first time, begin with web browsing (#1) and a search skill (#2). Those two cover the most common use case: "go find something and tell me about it."

From there, add based on your actual workflow. If you create presentations, add Felo Slides. If you need mobile access, add Telegram. If you run n8n automations, add the workflow skill.

The worst approach is installing 20 skills at once. Each skill adds context overhead, and your agent gets slower and more confused with every addition. Start small, add as needed.

Related Reading

• OpenClaw Skills Guide: A 2026 Developer's Guide
• Best OpenClaw Skills Worth Installing in 2026 (Plus What to Build With Each)
• SkillsMP Review 2026: What to Know, 66,500+ AI Agent Skills, and How to Choose