Connecting Infrastructure

Time to wire up your monitoring stack, cloud providers, and deployment tools to OpenClaw. Once connected, alerts and events flow into your chat automatically.

Prerequisites

Connect your monitoring platform

Grafana sends alerts via webhooks. Point them at OpenClaw:

openclaw connect add \
  --name "Grafana" \
  --type webhook \
  --secret "YOUR_WEBHOOK_SECRET" \
  --format grafana

This gives you a webhook URL. In Grafana:

Go to Alerting → Contact points → Add contact point
Choose Webhook as the type
Paste the OpenClaw webhook URL
Set the HTTP method to POST
Test the contact point

Copy webhook URL

openclaw connect show Grafana --url

Verify the connection:

Test connection

openclaw connect test Grafana

Connect your cloud infrastructure

For remediation actions (scaling, restarting, rolling back), OpenClaw needs access to your infrastructure:

openclaw connect add \
  --name "AWS Production" \
  --type aws \
  --region "us-east-1" \
  --access-key "YOUR_AWS_ACCESS_KEY" \
  --secret-key "YOUR_AWS_SECRET_KEY" \
  --capabilities "ec2,ecs,rds,lambda,autoscaling"

Connect deployment tools

Link your CI/CD pipeline so OpenClaw knows when deployments happen:

openclaw connect add \
  --name "GitHub" \
  --type github \
  --token "YOUR_GITHUB_PAT" \
  --repos "your-org/your-app" \
  --events "deployment,workflow_run"

Set up the monitoring agent

Create the base agent that receives and routes alerts:

openclaw agent add \
  --name "DevOps Monitor" \
  --source "Grafana,AWS Production,GitHub" \
  --message "You are a DevOps monitoring assistant. When you receive an alert or event:

1. Identify the severity: 🔴 Critical, 🟡 Warning, 🟢 Info
2. Summarize what happened in plain English
3. Add relevant context (recent deployments, related metrics)
4. For warnings and criticals, suggest immediate next steps

Keep messages concise. Engineers are reading this at 3 AM.
Don't use jargon the alert already contains — add *new* context." \
  --announce \
  --channel slack \
  --to "#ops-alerts"

Test it with a simulated alert:

openclaw connect simulate Grafana \
  --alert "CPU usage above 90%" \
  --metric "cpu_usage_percent" \
  --value 94 \
  --host "web-prod-01"

You should see a formatted alert in your chat channel.

What's Connected Now

  Monitoring                Cloud                  CI/CD
  ┌──────────┐           ┌──────────┐          ┌──────────┐
  │ Grafana  │           │ AWS/GCP  │          │ GitHub   │
  │ Datadog  │───────┐   │ or SSH   │──────┐   │ Vercel   │──┐
  │ CW/UR    │       │   └──────────┘      │   │ Docker   │  │
  └──────────┘       │                     │   └──────────┘  │
                     ▼                     ▼                 ▼
                  ┌──────────────────────────────────────────┐
                  │           OpenClaw Gateway               │
                  │    DevOps Monitor Agent                  │
                  └──────────────────────────────────────────┘
                                   │
                                   ▼
                          #ops-alerts channel

Raw alerts are now flowing to chat. The next lesson makes them smart — enriched with AI analysis, correlated with deployments, and filtered to reduce noise.

Knowledge Check

Why does OpenClaw need access to your deployment tools (GitHub/Vercel), not just monitoring?