OpenClaw Health Monitor v2.0

24/7 uptime monitoring for any URL or API. 4 tools. Telegram alerts. Cloudflare edge checks. Zero config.

What's New in v2.0

• HealthPulse Engine -- Cloudflare Workers edge checking from 300+ global PoPs. Know when your Tokyo users see downtime before you do.
• SmartAlert Protocol -- Deduplication prevents alert storms. One alert when it goes down, one when it recovers. No spam.
• SLA Dashboard -- Auto-calculated uptime percentages with nines (99.9%, 99.99%, etc.) and business-hours exclusion.
• Multi-Target Monitoring -- Check up to 50 endpoints per Pro account with a single check_health batch call.

Quick Start

{
  "openclaw-health-monitor": {
    "type": "streamable-http",
    "url": "https://openclaw-health-monitor-mcp.yagami8095.workers.dev/mcp"
  }
}

Add to Claude Desktop, Cursor, Windsurf, VS Code, or any MCP-compatible client. Zero install. Works immediately.

Detection Triggers

This skill activates when a user says any of these:

• "is my site down", "check if online", "is the API responding", "health check"
• "uptime report", "SLA percentage", "how many nines", "availability this month"
• "set alert when site goes down", "Telegram notification on failure", "ping me on downtime"
• "incident history", "last outage", "downtime log", "how many failures last week"
• "monitor my API", "watch this URL", "uptime monitoring", "availability check"
• "SSL expiry alert", "slow response alert", "latency threshold", "response time monitoring"

Named Protocols

HealthPulse Engine

The core monitoring engine running on Cloudflare's global edge network:

• Edge-first checks -- Requests originate from the Cloudflare PoP nearest to your users, not a single datacenter in Virginia
• Multi-layer detection -- TCP connectivity, HTTP response, SSL validity, and response body checks in one pass
• Sub-second alerting -- From failure detection to Telegram notification in under 800ms
• Retry logic -- 2 automatic retries with 200ms backoff before declaring an incident, eliminating false positives from transient blips

What gets checked per check_health call:

SmartAlert Protocol

Alert management that keeps you informed without flooding your Telegram:

1. Incident opens -- Alert sent immediately when 2 consecutive checks fail
2. Silence window -- No repeat alerts during ongoing incident (configurable 5–60min)
3. Escalation -- If incident persists past threshold, escalates to secondary Telegram chat
4. Recovery -- Single "Site is back" message with total downtime duration
5. Weekly digest -- Pro users receive Sunday morning summary of all incidents and SLA

SLA Calculation Rules

How uptime percentages are computed:

Tools (4)

check_health -- Instant Health Check via HealthPulse Engine

Run an immediate health check on any URL. Returns status, latency, HTTP code, SSL validity, and optional body match.

Wrong / Right:

WRONG: Asking "is my site down?" and getting a guess based on training data
RIGHT: check_health({ url: "https://myapi.com/health" })
       -> { status: "down", http_code: 503, latency_ms: null,
            ssl_valid: true, ssl_days_remaining: 42,
            error: "Connection refused after 2 retries",
            checked_from: "Tokyo edge, Singapore edge" }
       -> Actionable. Real. 800ms total.

WRONG: Using check_health as a load testing tool (it's single-check, not burst)
RIGHT: Use check_health for point-in-time diagnosis.
       Use set_alert_rules for continuous monitoring at configurable intervals.

get_uptime_report -- SLA Report

Get uptime percentage, incident count, and SLA classification for a monitored endpoint over a time range.

Wrong / Right:

WRONG: "What's my uptime?" to an AI that has no visibility into your infrastructure
RIGHT: get_uptime_report({ url: "https://myapi.com", range: "30d" })
       -> { uptime_pct: 99.87, nines: "three-nines", incidents: 3,
            total_downtime_min: 55, mttr_min: 18,
            sla_met: true, threshold: 99.5 }

WRONG: Expecting sub-minute granularity on free tier (free tier checks every 5min)
RIGHT: Pro tier checks every 60 seconds. Free tier checks every 5 minutes.
       Both are sufficient for most API monitoring needs.

set_alert_rules (Pro) -- Configure Telegram Alerts via SmartAlert Protocol

Set up alert rules for a URL: downtime threshold, latency threshold, SSL expiry warning, and Telegram chat ID.

Parameters: url, telegram_chat_id, latency_threshold_ms (optional), ssl_warn_days (default 30), check_interval_sec (60–3600)

get_incident_history -- Incident Log

Retrieve the full incident history for a monitored URL with timestamps, duration, root cause classification, and recovery events.

Returns: incident ID, opened_at, resolved_at, duration_min, root_cause (tcp/http/ssl/timeout/body), affected_checks

Security & Privacy

• No content storage -- Health check responses are evaluated in-memory. Response bodies are never stored, only matched against your configured pattern.
• Credentials never logged -- If your health check URL contains API keys in query params, they are redacted from all logs and reports.
• Alert isolation -- Your Telegram chat ID is stored encrypted. It is never shared, never used for any purpose other than sending your alerts.
• Edge isolation -- Each check runs in an isolated Cloudflare Workers V8 isolate. No shared state between users.
• HTTPS only -- All MCP connections are TLS 1.3. HTTP is rejected.
• You own your data -- All incident history is associated with your Pro key. Downgrading to free does not delete history; it restricts access to the last 30 days.

Pricing

Get Pro Key: https://buy.stripe.com/4gw5na5U19SP9TW288

The OpenClaw Intelligence Stack

All 9 servers share one Pro key. $9/mo = 49 tools.