What Are The Most Popular & Useful OpenClaw Skills: Top 20 With Use Cases

ClawHub, OpenClaw’s official skill registry, crossed 13,000 published skills in February 2026. The skills in this list were selected based on install counts, community star ratings, verified security status, and practical usefulness across real-world workflows, not just raw popularity. The top 20 below represent the skills that consistently appear across community recommendations, cover the widest range of use cases, and are safe to install on a properly hardened OpenClaw deployment.

Skills are what transform OpenClaw from a chatbot with file access into a genuinely useful automation platform. Without skills, your agent can read files, run shell commands, and respond to messages, but the workflows you can build are limited to whatever you explicitly program through prompts and config. Skills package pre-built capabilities (integrations, behaviors, search providers, output formats, scheduling patterns) into one-command installs that the agent can use immediately.

The single install command for any skill in this article:

Or from your connected chat interface:

How ClawHub Skills Work

Before the top 20, it is worth understanding what you are actually installing when you add a ClawHub skill, because this shapes how you evaluate them for security and capability. GrowExx’s top OpenClaw skills analysis describes ClawHub as the backbone of OpenClaw’s utility, where the difference between a basic agent and a genuinely productive one comes down almost entirely to which skills you install and how you configure them.

Every ClawHub skill is a Markdown-based package containing three core files:

• SKILL.md – Describes what the skill does, what triggers it, and what permissions it requires

• TOOLS.md – Defines the tools the skill exposes to the agent and what each tool actually executes

• README.md – Installation instructions, configuration options, and usage examples

The agent receives the skill’s tool definitions alongside its standard system prompt. When a task matches a skill’s trigger pattern, the agent invokes the skill’s defined tools rather than improvising raw shell commands. This makes skilled agents more predictable, more reliable, and easier to audit than agents that improvise everything from scratch.

Skills vary significantly in their permission requirements. Some require no external credentials at all (Capability Evolver, Self-Improving Agent, Mission Control). Others require OAuth connections (GOG for Google Workspace, Slack Supremacy). Others require third-party API keys (Tavily Search, ElevenLabs Agent). Always check the SKILL.md file before installing to understand exactly what access a skill requests.

The Top 20 OpenClaw Skills

1. Web Browsing

Install: openclaw skills install web-browsing
Install count: 180,000+ (most installed skill on ClawHub)
Security status: ✅ Bundled and verified
Requires: No API key

The Web Browsing skill is the single most essential install for any OpenClaw deployment. Without web access, the agent operates entirely on its training data, which has a knowledge cutoff and cannot access current events, live prices, documentation updates, or anything that has changed since the model was trained. Web Browsing gives the agent the ability to navigate URLs, extract structured content from pages, follow links, interact with web forms, and combine information from multiple pages within a single agent turn.

It is the foundation that every research-oriented workflow depends on. Virtually every other skill in this list works better with Web Browsing installed alongside it.

Primary use cases:

• Live fact-checking against current web sources before finalizing any output

• Research task: “Summarize the three most recent news articles about [topic]”

• Documentation lookups while debugging or developing

• Price monitoring: check competitor pricing pages on a schedule and alert if changes are detected

• Fetching and parsing structured data from web-accessible APIs or dashboards that do not have native integrations

• Pre-meeting research: “Look up [person/company] and give me background before my 2pm call”

Configuration note: Web Browsing uses OpenClaw’s native web_fetch tool under the hood. For agent-optimized search results (structured JSON, citation-ready output, agent-friendly parsing), install the Tavily skill (Skill #5) on top of Web Browsing. They complement each other well.

2. Capability Evolver

Install: openclaw skills install capability-evolver
Install count: 35,000+
Security status: ✅ Verified, bundled-tier trust
Requires: No API key

Capability Evolver is the most philosophically distinctive skill on ClawHub. Rather than adding a specific capability, it acts as a self-evolution engine that continuously improves how the agent handles your specific workflows over time. It monitors interaction patterns, identifies where the agent’s responses were suboptimal, and applies those learnings to future sessions, without requiring you to manually tune prompts or configs.

How it works technically:

• Analyzes session logs for patterns of correction (cases where you rewrote the agent’s output or asked it to redo something)

• Identifies recurring task types where response quality was below average

• Updates the agent’s working memory with refined handling instructions for those task types

• Runs the evolution cycle on a configurable schedule (typically overnight)

Primary use cases:

• Long-lived agents that handle the same recurring tasks weekly, where the agent gets measurably better at your specific workflow over months without manual tuning

• Teams running shared OpenClaw deployments where multiple users gradually shape the agent toward their collective preferences

• Any agent where you find yourself issuing the same correction more than twice, since Capability Evolver notices that pattern and adapts

Important note: Because Capability Evolver modifies the agent’s behavior over time, review its evolution logs periodically (openclaw skills logs capability-evolver) to confirm changes are moving in the direction you want.

3. Self-Improving Agent

Install: openclaw skills install self-improving-agent
Install count: 32,000+
Security status: ✅ Highest-starred skill on ClawHub (338 stars), no external dependencies
Requires: No API key

While Capability Evolver operates at the performance and pattern level, Self-Improving Agent operates at the knowledge and correction level. It captures every explicit correction, preference, and learned fact from your sessions into structured Markdown memory files. When you correct the agent (“I prefer bullet points over numbered lists,” “always use my first name,” “my timezone is Eastern”), the skill logs that learning immediately. Corrections that recur three or more times within 30 days are automatically promoted to permanent project memory.

This is the most starred skill on ClawHub by a wide margin, reflecting the genuine value of having an agent that remembers who you are and how you work.

Primary use cases:

• Building a personalized agent that adapts to your communication style over weeks of use

• Eliminating the need to repeat preferences in every new session

• Cross-session error correction: fix a mistake once, never see it again

• Building a persistent knowledge base about your projects, team, and context that carries across every conversation

Pair with: Capability Evolver. Together they cover both behavioral improvement (Capability Evolver) and explicit knowledge accumulation (Self-Improving Agent), creating an agent that improves along both dimensions simultaneously.

4. Mission Control

Install: openclaw skills install mission-control
Security status: ✅ Verified
Requires: Calendar and task management app access (configurable)

Mission Control is OpenClaw’s flagship productivity skill. It aggregates your calendar events, task lists, overdue items, and contextual priorities from recent work into a single structured daily briefing delivered at a scheduled time each morning. Users across the community consistently describe it as one of the most immediately impactful skills because it addresses the most universal daily friction: starting the day without a clear picture of what actually matters.

Mission Control triggers on the word briefing for on-demand use, but its real power is the cron-scheduled automatic morning delivery. This YouTube walkthrough of top OpenClaw use cases demonstrates Mission Control live in action alongside web research, MCP tools with Zapier, and content management workflows, making it the fastest way to see exactly how the skill behaves in a real daily setup.

Primary use cases:

• Daily 7AM briefing sent to Telegram or WhatsApp: overdue tasks, today’s calendar, priority context from yesterday’s work, and one proactive suggestion

• Pre-meeting prep: “Mission Control, what do I have in the next two hours and what background do I need?”

• End-of-day summary: “What did I complete today and what carries forward?”

• Weekly digest: Friday afternoon rollup of the week’s progress against goals

• Multi-project routing: when managing multiple workstreams, Mission Control surfaces which one is behind or at risk

Configuration example (agent chat):

5. Tavily Search

Install: openclaw skills install tavily
Security status: ✅ Verified, third-party API integration
Requires: Tavily API key (free tier: 1,000 searches/month)

Tavily is a search API built specifically for AI agents rather than human browsers, and it is one of the most widely recommended search providers in the OpenClaw community. Where generic web search returns HTML-heavy pages that the model has to parse and clean, Tavily returns structured JSON with pre-extracted content, clean text, citation metadata, and optional AI-synthesized answer summaries optimized for LLM consumption. It operates in two modes: basic (1 to 2 seconds, broad results) and advanced (5 to 10 seconds, deep content extraction).

Primary use cases:

• Deep research workflows where the agent needs to synthesize information from multiple sources into a coherent output

• Fact-checking: “Before publishing, verify these five claims against current sources”

• News monitoring: Tavily’s 7-day news-focused window returns recent articles for topics you want tracked

• Domain-scoped research: filter results to specific trusted domains (for example, only .gov or .edu sources for a regulatory question)

• Real-time data lookups: stock news, product launches, competitor announcements

Why it beats generic web search for agents: The structured JSON output from Tavily is significantly more reliable for agent workflows than HTML parsing of general web pages. Fewer tokens are wasted on navigation elements and ads, structure is more consistent for the model to reason about, and citation tracking is built in.

6. GOG (Google Workspace Integration)

Install: openclaw skills install gog
Install count: 14,000+
Security status: ⚠️ Review required (requires Google OAuth with broad workspace access)
Requires: Google account OAuth connection

GOG (Google Workspace) is the most popular productivity integration skill on ClawHub, providing a unified command-line-style interface to Gmail, Google Calendar, Google Drive, Google Docs, and Google Sheets. Once connected, the agent can read emails, create calendar events, draft and send messages, search Drive files, update spreadsheets, and create documents, all from natural language instructions in your messaging channel.

Primary use cases:

• Email triage: “Summarize my unread emails from the last 24 hours, flag anything urgent”

• Calendar management: “Block focus time every Tuesday and Thursday morning for the next month”

• Meeting prep: “Find the agenda doc for my 3pm meeting and summarize the key points”

• Spreadsheet automation: “Update the Q2 tracker sheet with this week’s numbers”

• Drive search: “Find all documents related to [project name] shared with me in the last 30 days”

• Draft and review: “Draft a reply to this email in a professional tone, then ask me to approve before sending”

Security note: GOG requires broad Google Workspace OAuth access because the tools span multiple Google services. Before installing, review the exact OAuth scopes the skill requests in its SKILL.md file. If you only need Gmail access, look for a Gmail-specific skill with narrower scopes. Always enable Human-in-the-Loop approval for any skill that can send emails or make calendar changes on your behalf.

7. Agent Browser

Install: openclaw skills install agent-browser
Install count: 11,000+
Security status: ✅ Verified
Requires: No API key (uses local browser automation)

Where Web Browsing fetches and reads static page content, Agent Browser provides full browser automation: clicking buttons, filling forms, scrolling pages, interacting with JavaScript-rendered content, navigating multi-step web flows, and extracting data from dynamic pages that require interaction to load. It uses a headless browser (Playwright) under the hood and gives the agent capabilities close to what a human user sitting at a keyboard can do in a browser.

Primary use cases:

• Multi-step form completion: filling out and submitting structured web forms

• E-commerce monitoring: tracking product availability and pricing changes on pages that block static scrapers

• Competitor intelligence: systematically visiting competitor pages, extracting structured data, and logging changes over time

• Account management tasks: log into a web dashboard, extract specific data, log out

• Web app testing: automated UI walkthrough of web applications

• Data extraction from dynamic pages: tables, dashboards, and content that only renders after JavaScript execution

Hardware note: Agent Browser is more resource-intensive than standard web fetching because it runs a full browser process. On a VPS or low-memory machine, configure it to run in non-persistent mode to avoid memory accumulation.

8. Deep Research

Install: openclaw skills install deep-research
Security status: ✅ Verified
Requires: Web Browsing and optionally Tavily for best results

Deep Research is a structured multi-step investigation skill that combines web search, content extraction, synthesis, and formatted reporting into a single triggered workflow. Unlike a standard web search that returns a summary of a few pages, Deep Research systematically explores a topic across many sources, cross-references findings, identifies contradictions, and produces a structured analytical report with citations. It is the closest thing to having a research assistant who does thorough work rather than summarizing the first three Google results.

Trigger keywords: research, deep research, analyze, audit, investigate

Primary use cases:

• Competitive landscape analysis: “Research the top five competitors to [product], compare their pricing, features, and reviews”

• Technology evaluation: “Deep research: what are the trade-offs between [Option A] and [Option B] for [use case]?”

• Due diligence: researching a company, person, or product before a meeting, investment, or partnership

• Academic-style topic research with citation tracking

• Market sizing: “Research the current market size, growth rate, and key players in [industry]”

• Technical documentation synthesis: combining information from multiple technical sources into a single coherent explanation

9. ElevenLabs Agent (Voice)

Install: openclaw skills install elevenlabs-agent
Security status: ✅ Verified, third-party API integration
Requires: ElevenLabs API key

The ElevenLabs Agent skill adds realistic voice output to your OpenClaw agent. Rather than receiving text responses in your messaging channel, the agent can generate and deliver spoken audio using ElevenLabs’ realistic voice synthesis. This transforms certain workflows from text-based to genuinely conversational and enables audio content creation entirely within OpenClaw.

Primary use cases:

• Voice briefings: have your morning Mission Control brief read aloud as an audio message to your phone while commuting

• Podcast content: generate scripted audio segments from research or draft content

• Accessibility workflows: convert long text documents to spoken audio for consumption during non-screen time

• Voice notifications: critical alerts and summaries delivered as voice messages rather than text

• Language learning: have the agent read text in a foreign language with native pronunciation

Cost note: ElevenLabs charges per character generated. Voice output is significantly more expensive than text-only workflows, so use it deliberately rather than enabling it for all outputs by default.

10. N8N Workflow Integration

Install: openclaw skills install n8n-automation
Security status: ✅ Verified (use webhook-only mode for production)
Requires: Running n8n instance

N8N is a workflow automation platform with 400+ native integrations. The N8N skill creates a two-way bridge: OpenClaw handles the AI reasoning and decision-making, while n8n handles the actual API calls and service integrations. As xCloud’s OpenClaw N8N integration guide documents, this pairing makes it possible to build secure AI agent pipelines that cleanly separate LLM reasoning from API execution, giving you full auditability at every step of a workflow.

Primary use cases:

• CRM automation: the agent classifies an inbound lead, n8n logs it to HubSpot or Salesforce with the correct fields

• Multi-channel notifications: agent triggers an n8n workflow that sends to Slack, email, and a Notion database simultaneously

• Invoice processing: agent extracts data from an uploaded PDF, n8n enters it into accounting software

• E-commerce operations: agent monitors inventory levels and triggers n8n workflows to reorder, notify suppliers, or update listings

• GitHub issue triage: agent classifies new issues by severity and type, n8n assigns labels, notifies the right team channel, and updates the project board

Security recommendation: In production, limit the N8N skill to triggering existing workflows via webhooks rather than granting it full API control to create or modify workflows. Webhook-only access gives you the integration capability with significantly lower security risk.

11. Clawflows (Workflow Orchestration)

Install: openclaw skills install clawflows
Security status: ✅ Verified
Requires: No API key

Clawflows is OpenClaw’s native multi-step workflow orchestration skill. Where other skills add individual capabilities, Clawflows lets you chain multiple skills and agent actions into structured, named workflows that execute in sequence. You define the workflow once in natural language, and Clawflows handles the sequencing, error handling, retry logic, and conditional branching.

Primary use cases:

• Content pipeline: research, draft, review, format, and publish, with each step using appropriate tools and the agent checking quality before proceeding

• Onboarding workflow: new team member trigger, create accounts, send welcome message, schedule intro meeting, add to relevant channels

• End-of-day wrap-up: log completed tasks, update project tracker, generate tomorrow’s priority list, send summary to Telegram

• Incident response: detect alert, classify severity, notify appropriate people, create tracking ticket, begin investigation steps

12. GitHub Integration

Install: openclaw skills install github
Security status: ✅ Verified, requires scoped GitHub token
Requires: GitHub Personal Access Token (use minimum required scopes)

The GitHub skill gives OpenClaw direct integration with GitHub repositories: reading issues, creating pull request comments, searching code, managing labels, tracking milestones, and monitoring repository activity. For developers using OpenClaw as a development workflow assistant, this is one of the highest-value integrations available.

Primary use cases:

• Issue triage: “Summarize all open issues tagged bug in [repo] and classify them by severity”

• PR review assistance: “Review this pull request description and check if it references the related issue”

• Release notes generation: “Generate release notes from all merged PRs since the last tag in [repo]”

• Codebase onboarding: “Explain the architecture of [repo] by reading the key files and documentation”

• Daily standup prep: “What GitHub activity happened across my watched repos yesterday?”

• Dependency monitoring: “Check if any packages in package.json have known vulnerabilities”

13. Email Management

Install: openclaw skills install email-management
Install count: Top 5 on Tencent Cloud’s 2026 ranking
Security status: ✅ Verified
Requires: Email account connection (IMAP/SMTP or OAuth)

Email Management is one of the most universally high-value OpenClaw skills because email is the one productivity tool almost every person uses but few have meaningfully automated. The skill handles reading, categorizing, drafting, and optionally sending emails through both IMAP/SMTP connections and OAuth integrations with Gmail and Outlook.

Primary use cases:

• Triage: classify incoming emails by urgency and category, surface only the ones that need your attention

• Draft generation: “Draft a polite follow-up email to the thread about [topic]”

• Automated responses: standard replies to recurring message types (FAQ responses, meeting confirmations)

• Newsletter monitoring: summarize newsletters from a specific sender rather than reading the full content

• Attachment processing: extract data from emailed attachments (PDFs, spreadsheets) and log key information

• Unsubscribe audit: “List all the mailing lists I’m on and draft unsubscribe requests for the ones I haven’t opened in 90 days”

Security note: Start with read-only mode (allow: ["read"] in tool permissions) before enabling send capabilities. The risk profile of an agent that can read your email is very different from one that can send it on your behalf.

14. Telegram Integration (Enhanced)

Install: openclaw skills install telegram-enhanced
Install count: Top 3 on Tencent Cloud’s 2026 ranking
Security status: ✅ Verified
Requires: Telegram Bot Token

OpenClaw ships with basic Telegram support natively. The Telegram Enhanced skill significantly expands what the agent can do within Telegram: inline keyboards, structured message formatting, file attachments, group chat management, scheduled messages, and read-access to group conversations for monitoring purposes.

Primary use cases:

• Interactive agent responses with inline buttons (Approve/Reject, Yes/No, Done/Defer)

• Sending formatted reports with tables and code blocks directly to Telegram

• Group chat monitoring: the agent reads a team Telegram group and surfaces action items or decisions requiring follow-up

• Scheduled Telegram messages: sending reports, reminders, or briefings at specified times

• File sharing: the agent generates a report as a PDF or CSV and sends it as a Telegram file attachment

15. Slack Supremacy

Install: openclaw skills install slack-supremacy
Security status: ✅ Verified
Requires: Slack OAuth token

For teams operating primarily in Slack, Slack Supremacy makes OpenClaw a first-class Slack participant. It handles proper threading (not just posting to channels), emoji reaction workflows, interactive message components, rich block formatting, file uploads, and channel management. The “supremacy” in the name refers to its depth of Slack integration compared to simpler Slack skills that only post text messages.

Primary use cases:

• Thread-aware responses: the agent replies in threads, not top-level channel messages

• Emoji-triggered workflows: reacting to a message with a specific emoji triggers an agent action (🔍 for research this, ✅ for mark as done, 📋 for create a task)

• Scheduled Slack digests: weekly roundup posts to team channels formatted with headers, bullet points, and links

• Interactive approval flows: the agent drafts a proposal and posts it with Approve/Request Changes buttons

• Channel summarization: “Summarize the activity in #product-feedback from the last 7 days”

16. Composio

Install: openclaw skills install composio
Security status: ✅ Verified, third-party managed integrations
Requires: Composio API key (free tier available)

Composio is an integration platform that gives OpenClaw access to over 150 managed API integrations through a single skill and a single API key. Rather than configuring individual OAuth connections for each service (Notion, Linear, Jira, Salesforce, HubSpot, Asana, Airtable, and many more), Composio manages the authentication and API translation, letting the agent interact with any supported service through natural language.

Primary use cases:

• Project management across tools: “Create a Linear issue from this bug report and link it to the Q2 roadmap item in Notion”

• CRM logging: “Log this call summary to HubSpot with the correct contact and deal association”

• Cross-tool data sync: “Update the Airtable tracker with the latest status from Jira for all open tickets”

• Notion workspace management: creating pages, updating databases, searching for documents

• Calendar integration beyond Google: Composio supports multiple calendar providers through a single interface

17. Finance Tracker

Install: openclaw skills install finance-tracker
Security status: ✅ Verified (run locally for sensitive financial data)
Requires: Bank export files or financial API connection

Finance Tracker gives OpenClaw the ability to process bank statements, expense records, invoices, and financial data files, categorizing transactions, identifying spending patterns, generating summaries, and flagging anomalies. It works with both manually uploaded export files (CSV from your bank) and API connections to financial data providers.

Primary use cases:

• Monthly spending summary: upload your bank export and get a categorized breakdown with comparisons to prior months

• Business expense categorization: classify receipts and transactions by project, client, or category for accounting purposes

• Invoice extraction: extract key data (amount, date, vendor, line items) from uploaded PDF invoices into a structured format

• Budget tracking: compare actual spending against defined budget categories and flag overruns

• Tax preparation assistance: categorize deductible business expenses from a year’s transaction history

Security note: Financial data is among the most sensitive possible input for an OpenClaw agent. Run this workflow locally with Ollama models rather than cloud API providers if privacy matters for your financial information.

18. Frontend Design Assistant

Install: openclaw skills install frontend-design
Security status: ✅ Verified
Requires: No API key

The Frontend Design skill equips OpenClaw with structured knowledge of UI/UX principles, component patterns, CSS/Tailwind conventions, accessibility standards, and responsive design best practices. It functions as a design-aware code generation assistant that produces frontend code with higher design quality than a generalist prompt. Felo AI’s 2026 ClawHub breakdown highlights Frontend Design as one of the most underrated skills on ClawHub, noting that developers who pair it with the GitHub skill get a near-complete development workflow assistant without any additional configuration.

Primary use cases:

• Component generation: “Build a responsive pricing table component in Tailwind CSS with three tiers”

• Design critique: “Review this landing page layout and identify UX problems”

• Accessibility audit: “Check this component for WCAG compliance issues and suggest fixes”

• Style system generation: “Create a complete color palette and typography scale based on this brand guide”

• Refactoring: “Refactor this CSS to Tailwind while preserving all visual styles”

19. Summarize

Install: openclaw skills install summarize
Security status: ✅ Verified, bundled-tier trust
Requires: No API key

Summarize is one of the simplest skills on this list, but its install count reflects how universally useful it is. It provides structured, format-aware summarization with configurable output templates: executive summaries, bullet points, topic-organized summaries, TLDR single-paragraph summaries, action-item extractions, and structured meeting notes.

Primary use cases:

• Meeting transcript to action items: paste or upload a meeting transcript, receive a structured list of decisions, owners, and deadlines

• Long document digest: summarize a lengthy PDF or article into the three most important points

• Email thread summary: what was decided and what is the current status of this thread?

• Research paper breakdown: key findings, methodology, limitations, and applicability to your context

• Changelog summarization: “Summarize what changed in this release compared to the last”

20. WaCLI (Utility Automation)

Install: openclaw skills install wacli
Security status: ✅ Verified
Requires: No API key

WaCLI is a Swiss-army-knife utility skill covering a wide range of small but high-frequency automation tasks that do not fit neatly into any single-purpose skill. It handles file format conversions, text transformations, date and time calculations, unit conversions, regex operations, clipboard workflows, and other utility operations that come up constantly in agent workflows but would be tedious to implement individually.

Primary use cases:

• File format conversion: “Convert this CSV to JSON”, “Convert this Markdown to HTML”

• Text transformations: clean formatting, case conversion, deduplication, line sorting

• Date and time utilities: “How many business days between these two dates?”, “Convert this timestamp to Eastern time”

• Batch file operations: rename, move, or reorganize files matching a pattern

• Data parsing: extract structured data from unstructured text using regex patterns

Complete Skill Stack Recommendations

Rather than installing all 20 skills at once, the community consensus is to build in tiers based on your use case:

The baseline stack (install these first):

The productivity stack (for daily automation):

The research stack (for information workers):

The developer stack:

ClawHub Stats at a Glance (2026)

Frequently Asked Questions

How do I find new skills beyond this list?
Browse ClawHub directly at clawhub.io or search from your chat interface: “search ClawHub for [topic] skills”. Filter by install count and star rating for a quick quality signal, and always check the publisher’s account history before installing anything.

Can I install multiple search skills (Tavily, Brave, Perplexity) simultaneously?
Yes. OpenClaw supports multiple search providers simultaneously and routes queries based on the skill’s trigger conditions and your configured preferences. Tavily handles deep research queries, Brave handles privacy-sensitive searches, and Perplexity handles synthesis-style queries where you want a cited AI-generated answer. They do not conflict.

Do skills work with local Ollama models?
Yes, with some caveats. Skills that require precise structured JSON output (like GitHub or GOG) work significantly better with larger, more capable models. With a 7B model like Mistral, simple skills (Summarize, WaCLI) work well, but complex multi-step skills (Deep Research, Clawflows) may produce unreliable results. A 32B+ model like Command R or Qwen2.5-Coder handles virtually the full skill library reliably.

Are skills free?
The skills themselves are free open-source packages. Some skills require third-party API keys that have their own pricing (Tavily, ElevenLabs, Composio). The majority of the top 20 in this list require no third-party credentials at all.

How do I uninstall a skill?

Or from your chat interface: “uninstall [skill-name] skill”. Uninstalling removes the skill’s tool definitions from the agent’s context and deletes its files from ~/.openclaw/skills/.

What is the difference between a skill and an MCP (Model Context Protocol) integration?
Skills are OpenClaw-native packages installed from ClawHub that follow the SKILL.md/TOOLS.md format. MCP integrations are a standardized protocol for connecting external tools and services to any compatible AI agent, not just OpenClaw. OpenClaw supports both. Skills are simpler and more OpenClaw-specific; MCP integrations are more portable and often have richer official support from the service provider (for example, official MCP servers from Notion, Linear, etc.).

Bottom Line

The twenty skills in this list collectively cover the full landscape of what makes OpenClaw genuinely useful in daily life: self-improvement and memory (Capability Evolver, Self-Improving Agent), daily productivity (Mission Control, GOG, Email Management), research and information gathering (Web Browsing, Tavily, Deep Research, Agent Browser), developer workflows (GitHub, Frontend Design, WaCLI), and cross-platform automation (N8N, Clawflows, Composio). Start with the baseline stack, add the research or productivity stack depending on your primary use case, and install individual skills from the rest of the list as specific needs arise.