Your agent has channels, personality, memory, rules, and skills. Now it needs deeper integrations — Google Workspace, web search backends, browser automation, webhooks, and custom integration chains.
WhatsApp, Telegram, Discord, Slack, iMessage, and more — all through one unified gateway.
Event-driven triggers from any external service. GitHub push, Stripe payment, form submission — any HTTP POST can wake your agent and pass it data.
Time-based automation. Cron for exact schedules, heartbeat for periodic check-ins. Your agent works on a clock, not just on demand.
Your agent’s logic is written once. The integration layer handles delivery. One brain, many mouths — the same agent can respond on WhatsApp, triage emails, and post to Slack without any per-channel customization.
External events trigger your agent via HTTP POST. GitHub pushes, Stripe payments, form submissions — anything that can send a webhook.
Schedule tasks at precise times. Daily reports at 8am, weekly summaries on Friday, monthly audits on the 1st. Standard cron syntax.
Periodic check-ins defined in HEARTBEAT.md. Your agent wakes itself on interval, runs its instructions, reports back.
Heartbeat for recurring agent-initiated work (briefings, monitoring). Cron for exact-time scheduling (reports, backups). Webhook for external event reactions (deploy notifications, payment alerts).
Email triage, drafting, sending, and inbox management. Your agent reads, categorizes, and responds to email on your behalf.
Schedule management, event creation, conflict detection, and meeting prep. Query availability, book time, send invites.
Document creation, file search, and content extraction. Your agent can create docs, find files, and pull data from your Drive.
Always use a dedicated Gmail account for your agent. Never connect your personal Gmail. The agent gets full read/write access — this is a security boundary you do not want to blur.
The gog skill handles OAuth authorization automatically. On first use, your agent walks you through granting access. Tokens are stored securely and refreshed as needed. No manual API key management required.
Install via openclaw skills install gog or tell your agent “Install the gog skill for Google Workspace.” The agent handles the rest.
Self-hosted, zero API costs, no rate limits. Aggregates results from multiple search engines. Full control over your search infrastructure.
Install by telling your agent “Install SearXNG” or configure manually in openclaw.json.
Simpler setup — just an API key. But comes with usage limits and costs per query. Good for getting started fast before migrating to SearXNG.
Set your search backend in openclaw.json or simply tell your agent which search provider to use. The agent configures it for you.
Your agent gets a full browser. It can navigate to URLs, fill forms, click buttons, take screenshots, extract data from rendered pages, and interact with JavaScript-heavy web apps that simple HTTP requests can’t handle.
Install Chromium via Playwright. The agent uses a dedicated browser profile — never your personal browser profile. This isolates cookies, sessions, and browsing history completely.
Browser automation blocks access to private networks by default — no browsing to localhost, 127.0.0.1, or internal IPs. This prevents Server-Side Request Forgery attacks where a prompt injection could make your agent probe your local network.
Need to access an internal service? Add specific hostnames to the allowlist as exceptions. This gives you controlled access without disabling SSRF protection entirely.
Write a SKILL.md with instructions, define parameters, build the handler. Full control over every detail of the integration.
Do something manually with your agent, then say “Turn what you just did into a skill.” The agent packages the workflow into a reusable Skill automatically.
Guide the agent step by step through the integration process. Teach it the API, the auth flow, the data format — it learns and codifies the workflow.
Install SearXNG (tell your agent “Install SearXNG”) or configure Brave Search API with your key. Test it by asking your agent a question about current events and verifying it searches the web to answer.
Install the gog skill and connect Gmail or Calendar using a dedicated account. Test with a simple task: “Check my calendar for tomorrow” or “Summarize my unread emails.”
Set up a cron job that runs a daily summary, or configure a webhook endpoint that your agent responds to. Test the trigger and verify the agent executes correctly.
Install Chromium via Playwright and ask your agent to browse a website, take a screenshot, or extract data from a page. Verify SSRF blocking is active by confirming it refuses to visit localhost.
Your agent is connected to the world. Now it needs a proper home. On Day 12 we containerize everything with Docker — isolated environments, reproducible deployments, and the foundation for running your agent in production safely and reliably.