Today is the watershed between toy and tool. You will:

• Understand OpenClaw's Skills system

• Connect Gmail — let your assistant read and send emails

• Connect Google Calendar — manage your schedule

• Configure web search — let your assistant find information online

• Unlock browser capabilities — let your assistant view any webpage

Over the past three days, your assistant already has a soul, a personality, and knows you. But it's still essentially a chat partner—you ask, it answers, that's it.

Today we're doing something game-changing: letting your assistant touch your real world.

Read emails. Check calendar. Search the web. Browse websites.

After today's configuration, when you tell your assistant "check what emails I have today," it can actually go check. Say "am I free tomorrow afternoon," it can actually check your calendar. Say "what's this product like," it can actually go search.

This is the watershed between toy and tool.

In OpenClaw, assistants gain new abilities through Skills. Each Skill is a set of configurations and scripts that tell the assistant how to use an external service.

Today we'll install four core skills:

This is your first "practical skill" and what most people need most.

1. Go to console.cloud.google.com

2. Create a new project (any name, like "My AI Assistant")

3. Go to APIs & Services > Library, search and enable:

   • Gmail API

   • Google Calendar API

1. Go to APIs & Services > Credentials

2. Click Create Credentials > OAuth client ID

3. Application type: choose Desktop app

4. Download the JSON file, name it credentials.json

MyClaw Cloud: Upload your credentials.json through the Dashboard file manager into your instance's workspace directory. Then use the built-in skill installer in the Dashboard to install the gog (Google Workspace) skill, which includes Gmail + Google Calendar + Google Drive. The Dashboard will guide you through the OAuth authorization flow.

Self-hosted: Place credentials.json in your working directory (~/clawd/credentials.json), then install the skill:

gog is the Google Workspace skill, which includes Gmail + Google Calendar + Google Drive.

The first time you run it, it will open a browser link for you to authorize your Google account. After authorization, it generates a token.json—that's your key.

Tell your assistant in Telegram:

Check what new emails I have in Gmail today

If everything is working, you'll get a response like this:

5 new emails today:

1. [Important] Partner reply — About next week's meeting time confirmation, needs your response

2. GitHub — Your repository has been starred

3. Cloud Provider — Invoice for July

4. Newsletter — This Week in AI

5. Ads — Automatically ignored

Notice that? It doesn't just list emails, it helps you judge priority and flags the one that needs your attention. That's the difference between an AI assistant and a regular email client.

With Gmail set up, calendar is simple—they share the same Google OAuth authentication.

Since you already authorized your Google account when installing the gog skill, and you enabled the Calendar API in Step 1, calendar functionality works directly without extra steps.

Test it:

What do I have tomorrow?

Tomorrow's schedule (Saturday):

• 10:00-11:00 Product Discussion (Video Call)

• 14:30 Dentist appointment

• No other events, afternoon free for deep work

More powerful usage:

Create a meeting for next Wednesday at 3 PM, topic "SEO Strategy Discussion," duration 1 hour

Calendar event created:

• Wednesday 15:00-16:00

• SEO Strategy Discussion

• Want to add attendees?

It can even detect conflicts—if the time slot you want is already taken, it'll alert you:

Note: You already have "Client call" scheduled for Wednesday 15:00-16:00. Should I make it start at 16:30 instead?

Before, you'd open your calendar app, scroll up and down to find free slots, manually create events. Now? One sentence.

Letting your assistant search the web is key to breaking the "information silo."

OpenClaw supports multiple search methods. The simplest is Brave Search API:

1. Go to brave.com/search/api and register a free account

2. Get your API Key

MyClaw Cloud: Add the Brave Search API Key through the Dashboard's skill configuration panel. Navigate to Settings > Skills > Web Search and enter your key.

Self-hosted: Add it to OpenClaw configuration:

openclaw configure --section web

The wizard will prompt you to enter your Brave Search API Key and automatically write it to the config.

After configuration, test it:

Search "OpenClaw alternatives 2026"

Search results summary: Main alternatives to OpenClaw include:

1. AgentGPT — Runs in browser, no deployment needed

2. AutoGPT — Classic Agent project, large community

3. CrewAI — Multi-Agent collaboration framework

But these focus on "autonomous task execution," while OpenClaw focuses on "personal assistant"—24/7 online, chat interaction, skills system. Different positioning, not direct competitors.

It doesn't just dump search results at you—it reads through them, summarizes, and gives you judgment. That's something search engines can't do.

Some information search engines can't find—like specific content on a particular webpage, data from a dashboard requiring login, a dynamically loaded page.

That's when you need the browser skill—letting your assistant "see" webpages.

OpenClaw has a built-in browser skill (based on Playwright), already auto-configured during installation. It can:

• Visit any URL and extract content

• Take screenshots of the current page

• Interact with clicks, inputs, scrolling

Usage example:

Open https://example.com and show me what the homepage looks like now

Visited example.com:

• Homepage title: "Example — Free Online Tools"

• Main sections: Features, Pricing, Blog

• Page loaded normally, no visible errors [Screenshot saved]

More practical scenario:

Check competitor xyz.com's pricing page

It will open the page, extract pricing information, and even compare with previous versions you've seen.

With email, calendar, and browser connected—your assistant can now touch a lot of personal data. Security is something you must take seriously.

MyClaw Cloud: Your instance runs in an isolated environment with enterprise-grade security. However, you still need to manage your own API keys and OAuth tokens carefully. The Dashboard provides a secure credential store for managing these.

Self-hosted: I recommend running a security check:

openclaw security audit
openclaw security audit --deep

1. API Key Security

• Never commit API Keys to Git

• Store in environment variables or .env files

• Rotate keys regularly

2. OAuth Token Security

• Files like token.json contain your Google authorization info

• Make sure file permissions are set correctly: chmod 600 token.json

• Don't upload to any public place

3. Principle of Least Privilege Only give your assistant the permissions it needs. For Gmail, if you only need to read emails, don't give "send email" permission. Although OpenClaw requires confirmation before sending by default, one fewer permission means one fewer risk.

4. Behavioral Boundaries Clearly write in SOUL.md and AGENTS.md:

• What operations need confirmation

• What data cannot be externally shared

• When to refuse execution

Security isn't a one-time thing—it's an ongoing habit. API Keys don't go in repos, Token files need proper permissions, least privilege principle, behavioral boundaries clearly written.

• Skills system: Skills are how your assistant gains new abilities, like installing phone apps

• Gmail connection: gog skill + OAuth authorization, assistant can read/send emails

• Calendar connection: Same gog skill, assistant can view and manage your schedule

• Search capability: Brave Search API lets your assistant find information online

• Browser capability: Let your assistant "see" and interact with webpages

• Security first: API Keys don't go in repos, least privilege, clear behavioral boundaries

Today was a "capability explosion" day:

• Connected Gmail — assistant can read your emails now

• Connected Google Calendar — assistant can manage your schedule now

• Configured search engine — assistant can find information online now

• Enabled browser skill — assistant can "see" webpages now

• Built security awareness — know how to protect your data

From today, your assistant is no longer a toy that can only chat—it's a tool that can actually help you get things done.

Try telling it: "Check what emails I have today, what I have scheduled tomorrow, and search for recent AI news."

One sentence, three things, all handled. Before, that meant opening three apps, spending ten minutes. Now? Ten seconds.

Gmail and calendar are just the beginning. OpenClaw has a complete Skills ecosystem—SEO analysis, social media management, code review, PDF parsing, database queries... Tomorrow we'll browse the skill marketplace and arm your assistant to the teeth.

Today is the watershed between toy and tool. You will:

• Understand OpenClaw's Skills system

• Connect Gmail — let your assistant read and send emails

• Connect Google Calendar — manage your schedule

• Configure web search — let your assistant find information online

• Unlock browser capabilities — let your assistant view any webpage

Over the past three days, your assistant already has a soul, a personality, and knows you. But it's still essentially a chat partner—you ask, it answers, that's it.

Today we're doing something game-changing: letting your assistant touch your real world.

Read emails. Check calendar. Search the web. Browse websites.

After today's configuration, when you tell your assistant "check what emails I have today," it can actually go check. Say "am I free tomorrow afternoon," it can actually check your calendar. Say "what's this product like," it can actually go search.

This is the watershed between toy and tool.

In OpenClaw, assistants gain new abilities through Skills. Each Skill is a set of configurations and scripts that tell the assistant how to use an external service.

Today we'll install four core skills:

This is your first "practical skill" and what most people need most.

1. Go to console.cloud.google.com

2. Create a new project (any name, like "My AI Assistant")

3. Go to APIs & Services > Library, search and enable:

   • Gmail API

   • Google Calendar API

1. Go to APIs & Services > Credentials

2. Click Create Credentials > OAuth client ID

3. Application type: choose Desktop app

4. Download the JSON file, name it credentials.json

MyClaw Cloud: Upload your credentials.json through the Dashboard file manager into your instance's workspace directory. Then use the built-in skill installer in the Dashboard to install the gog (Google Workspace) skill, which includes Gmail + Google Calendar + Google Drive. The Dashboard will guide you through the OAuth authorization flow.

Self-hosted: Place credentials.json in your working directory (~/clawd/credentials.json), then install the skill:

gog is the Google Workspace skill, which includes Gmail + Google Calendar + Google Drive.

The first time you run it, it will open a browser link for you to authorize your Google account. After authorization, it generates a token.json—that's your key.

Tell your assistant in Telegram:

Check what new emails I have in Gmail today

If everything is working, you'll get a response like this:

5 new emails today:

1. [Important] Partner reply — About next week's meeting time confirmation, needs your response

2. GitHub — Your repository has been starred

3. Cloud Provider — Invoice for July

4. Newsletter — This Week in AI

5. Ads — Automatically ignored

Notice that? It doesn't just list emails, it helps you judge priority and flags the one that needs your attention. That's the difference between an AI assistant and a regular email client.

With Gmail set up, calendar is simple—they share the same Google OAuth authentication.

Since you already authorized your Google account when installing the gog skill, and you enabled the Calendar API in Step 1, calendar functionality works directly without extra steps.

Test it:

What do I have tomorrow?

Tomorrow's schedule (Saturday):

• 10:00-11:00 Product Discussion (Video Call)

• 14:30 Dentist appointment

• No other events, afternoon free for deep work

More powerful usage:

Create a meeting for next Wednesday at 3 PM, topic "SEO Strategy Discussion," duration 1 hour

Calendar event created:

• Wednesday 15:00-16:00

• SEO Strategy Discussion

• Want to add attendees?

It can even detect conflicts—if the time slot you want is already taken, it'll alert you:

Note: You already have "Client call" scheduled for Wednesday 15:00-16:00. Should I make it start at 16:30 instead?

Before, you'd open your calendar app, scroll up and down to find free slots, manually create events. Now? One sentence.

Letting your assistant search the web is key to breaking the "information silo."

OpenClaw supports multiple search methods. The simplest is Brave Search API:

1. Go to brave.com/search/api and register a free account

2. Get your API Key

MyClaw Cloud: Add the Brave Search API Key through the Dashboard's skill configuration panel. Navigate to Settings > Skills > Web Search and enter your key.

Self-hosted: Add it to OpenClaw configuration:

openclaw configure --section web

The wizard will prompt you to enter your Brave Search API Key and automatically write it to the config.

After configuration, test it:

Search "OpenClaw alternatives 2026"

Search results summary: Main alternatives to OpenClaw include:

1. AgentGPT — Runs in browser, no deployment needed

2. AutoGPT — Classic Agent project, large community

3. CrewAI — Multi-Agent collaboration framework

But these focus on "autonomous task execution," while OpenClaw focuses on "personal assistant"—24/7 online, chat interaction, skills system. Different positioning, not direct competitors.

It doesn't just dump search results at you—it reads through them, summarizes, and gives you judgment. That's something search engines can't do.

Some information search engines can't find—like specific content on a particular webpage, data from a dashboard requiring login, a dynamically loaded page.

That's when you need the browser skill—letting your assistant "see" webpages.

OpenClaw has a built-in browser skill (based on Playwright), already auto-configured during installation. It can:

• Visit any URL and extract content

• Take screenshots of the current page

• Interact with clicks, inputs, scrolling

Usage example:

Open https://example.com and show me what the homepage looks like now

Visited example.com:

• Homepage title: "Example — Free Online Tools"

• Main sections: Features, Pricing, Blog

• Page loaded normally, no visible errors [Screenshot saved]

More practical scenario:

Check competitor xyz.com's pricing page

It will open the page, extract pricing information, and even compare with previous versions you've seen.

With email, calendar, and browser connected—your assistant can now touch a lot of personal data. Security is something you must take seriously.

MyClaw Cloud: Your instance runs in an isolated environment with enterprise-grade security. However, you still need to manage your own API keys and OAuth tokens carefully. The Dashboard provides a secure credential store for managing these.

Self-hosted: I recommend running a security check:

openclaw security audit
openclaw security audit --deep

1. API Key Security

• Never commit API Keys to Git

• Store in environment variables or .env files

• Rotate keys regularly

2. OAuth Token Security

• Files like token.json contain your Google authorization info

• Make sure file permissions are set correctly: chmod 600 token.json

• Don't upload to any public place

3. Principle of Least Privilege Only give your assistant the permissions it needs. For Gmail, if you only need to read emails, don't give "send email" permission. Although OpenClaw requires confirmation before sending by default, one fewer permission means one fewer risk.

4. Behavioral Boundaries Clearly write in SOUL.md and AGENTS.md:

• What operations need confirmation

• What data cannot be externally shared

• When to refuse execution

Security isn't a one-time thing—it's an ongoing habit. API Keys don't go in repos, Token files need proper permissions, least privilege principle, behavioral boundaries clearly written.

• Skills system: Skills are how your assistant gains new abilities, like installing phone apps

• Gmail connection: gog skill + OAuth authorization, assistant can read/send emails

• Calendar connection: Same gog skill, assistant can view and manage your schedule

• Search capability: Brave Search API lets your assistant find information online

• Browser capability: Let your assistant "see" and interact with webpages

• Security first: API Keys don't go in repos, least privilege, clear behavioral boundaries

Today was a "capability explosion" day:

• Connected Gmail — assistant can read your emails now

• Connected Google Calendar — assistant can manage your schedule now

• Configured search engine — assistant can find information online now

• Enabled browser skill — assistant can "see" webpages now

• Built security awareness — know how to protect your data

From today, your assistant is no longer a toy that can only chat—it's a tool that can actually help you get things done.

Try telling it: "Check what emails I have today, what I have scheduled tomorrow, and search for recent AI news."

One sentence, three things, all handled. Before, that meant opening three apps, spending ten minutes. Now? Ten seconds.

Gmail and calendar are just the beginning. OpenClaw has a complete Skills ecosystem—SEO analysis, social media management, code review, PDF parsing, database queries... Tomorrow we'll browse the skill marketplace and arm your assistant to the teeth.