What are OpenClaw skills?

OpenClaw skills are installable packages that add reusable capabilities to OpenClaw. Each skill contains instructions for a specific tool, workflow, or task, allowing the same behavior to be reused across multiple agent sessions.

Skills work alongside OpenClaw agents. The agent provides the core environment for running tasks, while installed skills give the agent task-specific instructions it can apply when relevant.

This helps the agent handle familiar tools, workflows, or task types more consistently.

Examples of skill types for common workflows:

• Browser automation for website navigation, form filling, and data extraction.
• Research for source discovery, fact-checking, and structured reports.
• Email and inbox management for reply drafting, thread summaries, and inbox organization.
• Calendar workflows for scheduling, availability checks, and meeting coordination.
• Lead management for prospect research, CRM updates, and outreach.
• Customer support for ticket triage, response drafting, and escalations.
• Coding and development for code review, debugging, and testing.
• Social media planning for content calendars, post creation, and campaign planning.
• File and document workflows for document summaries, information extraction, and report generation.
• Communication tool integrations for Slack, Discord, CRM, and project management workflows.

How OpenClaw skills work

OpenClaw skills work as structured instruction packages that the agent can load and apply during a task.

Each skill defines what the skill is for, when the agent should use it, and which steps or rules the agent should follow.

The main file in a skill is <code>SKILL.md</code> For Git and local installs, OpenClaw expects this file at the source root. 

The file contains the skill definition, including frontmatter that can identify the skill name. If the frontmatter name is valid, OpenClaw uses it during installation. If not, OpenClaw falls back to the source folder or repository name.

Once installed, a skill becomes available in the workspace where it was added. Installing with <code>–global</code> places the skill in the shared managed skills directory, so multiple workspaces can use the same skill.

During agent work, the skill acts as a reusable context. It can tell the agent how to use a tool, follow a workflow, reference files, format an output, or interact with an external service.

The agent still runs inside OpenClaw’s core environment, while the skill adds task-specific guidance that can be installed, updated, inspected, or removed separately.

For example, you can search for a published skill:

openclaw skills search "calendar"

Or install a ClawHub skill:

openclaw skills install @owner/<slug>

10 OpenClaw skill types

OpenClaw skills support a wide range of workflows, from browser automation and software development to research, document processing, customer support, and business automation.

The categories listed cover some of the most popular OpenClaw use cases and the types of skills available for each one.

1. Browser automation skills

Browser automation skills help agents interact with websites that require a real browser.

Common tasks include navigating pages, filling forms, checking dashboards, extracting data from JavaScript-rendered sites, taking screenshots, and automating multi-step workflows.

A good starting point is Playwright Browser Automation (@spiceman161/playwright-browser-automation), which provides browser control through the Playwright API for navigation, form interaction, screenshots, PDF generation, data extraction, and workflow automation.

If your work depends on rendered pages or browser-driven testing, Playwright Skill (@vmercel/playwright-skill) focuses on testing websites, validating login flows, checking responsive layouts, and generating Playwright scripts for custom automation.

For web scraping, Playwright Scraper (@3coco3/playwright-scraper) adds browser automation with JavaScript rendering and stealth features for sites that rely on dynamic content or anti-bot protection.

2. Research skills

Research skills guide agents through source discovery, information gathering, and structured synthesis.

They are useful for competitor research, market research, fact-checking, summaries, and reports that need clear source trails.

You can start with Deep Research (@ivangdavila/in-depth-research). It supports multi-source investigation with methodology tracking, source evaluation, and different research depth levels.

For academic or citation-heavy work, Academic Deep Research Pro (@nancliu/academic-deep-research-pro) focuses on structured research cycles, evidence hierarchy, APA 7th citations, and review checkpoints.

Web Researcher (@aokikaito79/web-researcher) is useful for current information and technical news because it guides the agent through multi-query searches, source checks, and fact verification before producing a summary.

Self-hosted search workflows can use WebSearch-Openclaw (@mr-ds-ml-85/searxng-websearch), which connects research tasks to a SearXNG instance for web search, page fetching, and online information gathering.

3. Email and inbox workflow skills

Email and inbox workflow skills structure recurring message tasks. They cover reply drafting, thread summaries, follow-up preparation, inbox triage, label updates, and message organization.

Email (@awspace/email-skill) covers broad email management across multiple providers, including sending, reading, searching, organizing messages, and handling attachments.

If your inbox runs on Gmail, gogcli-mcp-gmail (@chrischall/gogcli-mcp-gmail) supports deeper mailbox tasks such as working with threads, labels, drafts, attachments, forwarding, bulk archiving, trash, and mark-read operations.

Teams that need triage can use Inbox Triage (@clawdssen/inbox-triage) to scan messages, categorize urgency, draft replies, surface important items, and archive low-value email.

Microsoft 365 users may prefer Outlook Inbox (@hith3sh/outlook-inbox), which works with Outlook mail through Microsoft Graph and supports searching mail, reading threads, managing drafts, sending replies, and coordinating calendar events.

Inbox skills can read and draft sensitive messages, so make sure to review generated replies before sending, and confirm that each skill has only the mailbox access needed for the task.

4. Calendar skills

Calendar skills help manage time-based work. Use them to schedule meetings, check availability, prepare agendas, update event details, and coordinate follow-ups.

Google Calendar (@adrianmiller99/google-calendar) connects OpenClaw to the Google Calendar API for listing events, creating meetings, updating details, and deleting calendar entries.

If you want broader calendar management through natural language, Claw Calendar (@5twang/claw-calendar) supports creating, viewing, updating, and deleting events through the Claw Calendar API.

Meeting Scheduler Pro (@nollio/normieclaw-meeting-scheduler-pro) is useful for teams that coordinate frequent meetings, since it supports scheduling, availability checks, preparation briefs, agendas, and follow-up tasks.

5. Lead management skills

Lead management skills support prospect research, qualification, CRM preparation, outreach drafts, and follow-up tracking.

Use them when sales work needs a repeatable process from finding potential customers to recording the next action.

B2B Lead Generation (@ericn26-star/eric-b2b-lead-generation) helps identify prospects from product descriptions, find competitor references, prepare outreach material, and organize lead research into structured reports.

LinkedIn Lead Gen Outreach (@gaelbuenobarthe/linkedin-lead-gen-outreach) focuses on review-first LinkedIn prospecting, including lead quality checks, concise outreach messaging, and export-ready sales notes.

HubSpot API (@byungkyu/hubspot-api) connects OpenClaw to HubSpot for working with contacts, companies, deals, and other CRM objects through the HubSpot API.

If your sales pipeline runs on Salesforce, Salesforce API (@byungkyu/salesforce-api) lets agents retrieve and update CRM records through authenticated API calls.

6. Customer support skills

Customer support skills can help classify tickets, summarize customer problems, draft replies, prepare escalation notes, and suggest follow-up steps.

Zendesk (@ivangdavila/zendesk) connects agents to Zendesk for ticket lookup, ticket updates, comments, user records, and support workflow management.

If your team works from Intercom, Intercom (@georgelewi5/intercom-docs) supports customer lookup, conversation management, contact updates, and support messaging tasks.

Freshdesk (@hith3sh/freshdesk-support) is useful for teams that manage tickets, contacts, companies, and support notes inside Freshdesk.

7. Coding and development skills

Coding and development skills support code review, debugging, documentation updates, test generation, repository navigation, and development workflow assistance.

Code Review (@wpank/code-review) gives the agent a structured review process for pull requests and code changes. It covers security, performance, maintainability, correctness, and testing, with severity levels, feedback guidance, common review steps, and anti-patterns to avoid.

When you’re tracking down a bug, Code Debug (@sf0799/code-debug) helps diagnose errors and crashes in local code, identify the root cause, recommend the smallest effective fix, and outline verification steps to confirm the issue has been resolved.

8. Social media planning skills

Social media planning skills help you create content calendars, draft posts, repurpose long-form content, plan campaigns, and prepare publishing workflows.

Content Calendar (@chris-openclaw/content-calendar-os) helps plan, create, and track content across platforms, including topic brainstorming, post drafts, and content repurposing.

Social Media Scheduler (@mkpareek0315/social-media-planner) focuses on captions, hashtags, content calendars, and posting schedules. It does not publish posts directly, so users copy and post the generated content themselves.

Social Media Manager (@tryan310/social-media-manager) automates AI-generated social content and draft scheduling through the Postiz API. It supports multi-platform workflows for TikTok, Instagram, X/Twitter, LinkedIn, Facebook, YouTube, and other platforms connected in Postiz.

Social Media Autopilot (@reighlan/social-media-autopilot) supports multi-platform post scheduling, content calendars, approval workflows, engagement analytics, and brand voice consistency.

9. File and document workflow skills

File and document workflow skills are valuable add-ons for summarizing files, extracting key details, comparing documents, preparing reports, and turning unstructured content into organized findings.

Document Skill (@mskrq/document-skill) supports PDF, DOCX, PPT, and WPS document parsing, including key information extraction, table extraction, summaries, and cross-document content merging.

File Summary & Analysis (@leonthepro2012/file-summary) focuses on summarizing local files, including PDF, DOCX, XLSX, XLS, and TXT files.

If you need PDF-specific handling, pdf-skill (@weaglewang/pdf-skill) supports PDF reading, editing, merging, splitting, text extraction, table extraction, form filling, OCR, and HTML-to-PDF conversion.

Doc (@flywhale-666/doc) is useful when DOCX layout matters, since it focuses on reading, creating, editing, and visually checking Word documents with tables, diagrams, and pagination.

10. Communication tool integration skills

Communication tool integration skills help with summarizing conversations, sending status updates, routing tasks, managing channels, and connecting agent work to platforms such as Slack, Discord, or project management systems.

Slack (@hith3sh/slack-workspace) supports Slack workspace tasks such as sending messages, managing channels, searching conversations, handling files, creating reminders, and automating team communication.

For webhook-based Slack messages, slack-automator (@xueyetianya/slack-automator) supports channel notifications, direct messages, scheduled messages, templates, formatting helpers, and send history export.

Discord (@codedao12/discord-hub) covers Discord Bot API workflows, including interactions, commands, messages, and server operations through HTTPS requests.

Why users install or create OpenClaw skills

Users install or create OpenClaw skills when the same task needs to be handled the same way more than once.

A skill turns repeated instructions into a reusable package, so the agent can follow a defined process without requiring the user to explain every step again.

The main benefits include:

• Less repeated prompting – users can save task instructions, tool context, formatting rules, and workflow steps inside a skill.
• More consistent task execution – agents can follow the same process for recurring work, such as code review, research, support triage, reporting, or document analysis.
• Tool-specific guidance – skills can describe how an agent should work with a particular tool, service, file type, API, or workflow.
• Reusable team processes – teams can package internal standards, checklists, escalation rules, review criteria, and output formats as skills.
• Easier workspace setup – users can install the same skill across workspaces instead of rebuilding the same workflow manually.
• Cleaner agent behavior – skills keep task-specific instructions separate from the agent’s core environment, making capabilities easier to add, inspect, update, or remove.
• Safer skill creation flow – Skill Workshop gives users a CLI-based way to propose, review, apply, reject, or quarantine skill proposals before adding them to active workflows.

Safety tips before using OpenClaw skills

OpenClaw security starts with understanding what each skill can access.

Warning! Connecting an account grants the skill access to everything within that account’s scope. Avoid using admin or owner-level credentials. Use a dedicated service account or the minimum permission level required for the task.

Before installing or running a skill, review how it works, what it can access, and where it comes from:

• Read <code>SKILL.md</code> first – confirm what the skill does, when the agent should use it, which tools it references, and what output format it expects.
• Check access before connecting accounts – if a skill uses an inbox, calendar, CRM, API, browser, shell command, local file, or Git repository, review the requested permissions and connect only the account or workspace needed for the task.
• Inspect file and command behavior – avoid skills that ask the agent to run broad commands, modify large directories, or access sensitive paths without explaining why.
• Prefer trusted sources – install skills from publishers, repositories, or teams you recognize. Skip packages with unclear ownership, vague descriptions, or instructions that do not match the stated purpose.
• Use ClawHub verification when appropriate – OpenClaw checks community ClawHub installs and updates before downloading them. Malicious or blocked community releases are refused.
• Review risky releases manually – risky community releases require explicit acknowledgement before non-interactive installs can continue, so inspect the skill before approving it.
• Test with low-risk data first – run the skill on sample files, test workspaces, or non-sensitive tasks before using it with production data, customer records, or private repositories.

How to choose the right OpenClaw skill

The right OpenClaw skill depends on the task you want the agent to handle, the tool involved, and the level of access the workflow requires.

Start by identifying the work pattern, then choose the narrowest skill type that matches it.

Choose a browser automation skill for tasks that happen inside websites or dashboards, such as form filling, page navigation, screenshot capture, browser testing, or data extraction from JavaScript-rendered pages.

A research skill fits work that depends on source discovery, comparison, fact-checking, or structured findings.

Use this category for competitor research, market summaries, technical research, and reports that need a clear source trail.

For messages, meetings, reminders, and scheduling, choose an email or calendar skill. These skills fit inbox triage, reply drafts, thread summaries, availability checks, agenda preparation, and follow-up planning.

A CRM or lead management skill works best when the workflow involves prospects, contacts, deals, outreach, or sales follow-ups.

Choose this category for lead research, qualification, CRM record updates, and pipeline preparation.

Customer issues belong to a customer support skill. This category fits ticket classification, issue summaries, response drafts, escalation notes, and next-step suggestions.

Pick a coding skill when the agent needs to inspect, review, debug, document, or test code. These skills support code review, debugging, repository navigation, test generation, documentation updates, and assistance with development workflows.

A document workflow skill is the right fit when the task depends on files. Use it for PDF or DOCX summaries, table extraction, document comparison, report preparation, and structured outputs from unstructured content.

Choose a communication integration skill when work needs to move through Slack, Discord, project management tools, or team channels. These skills support status updates, conversation summaries, task routing, and team notifications.

OpenClaw best practices for using skills focus on improving an agent’s capabilities without making its behavior harder to understand, review, or control.

How to start using OpenClaw skills

Setting up OpenClaw is the first step before you can use the CLI to search, install, verify, update, or manage skills.

OpenClaw provides the agent environment, while the skills system stores and loads reusable instructions from supported sources.

Follow this setup flow to add a skill safely and confirm that the right agent can use it:

1. Set up OpenClaw – deploy an OpenClaw instance and confirm that the CLI can access the workspace where your agents will run.
2. Search for a relevant skill – use the OpenClaw CLI to find skills that match the task, tool, or workflow you want to add.
3. Review the source – inspect the skill’s <code>SKILL.md</code> file, owner, repository, required tools, and any referenced files or services before installation.
4. Install the skill – install from ClawHub, a Git repository, or a local directory. Git and local installs must include <code>SKILL.md</code> at the source root.
5. Inspect installed skills – list or inspect skills after installation to confirm the name, source, and installation location.
6. Check agent visibility – confirm whether the skill is installed in the current workspace or globally with <code>–global</code>, then verify that the intended agent can access it.
7. Run a controlled test – use a sample task, test repository, or non-sensitive file before letting the skill work with production data, customer records, or connected accounts.

If you’re comparing top OpenClaw hosting providers, look for a platform that gets OpenClaw running quickly without requiring manual installation or infrastructure management. 

Hostinger Managed OpenClaw provides a managed, one-click deployment option. It automates the environment setup, applies security patches, keeps OpenClaw on the latest stable version, and handles backups and infrastructure, so you can start exploring skills sooner.

All of the tutorial content on this website is subject to

Hostinger’s rigorous editorial standards and values.