Monitors · Logs · MCP server, on every plan

Your app breaks. Your AI assistant already knows why.

UptimeEye runs your real user flows every 30 seconds and keeps your logs next to every alert. Connect Claude Code or Cursor over MCP and the agent that wrote the code can read the alert, search the logs and fix the cause.

Start freeConnect your assistant in one line

14 days of Pro included, then free forever · no credit card · commercial use allowed.

  • Claude Code
  • Cursor
  • VS Code
  • Claude Desktop
  • any MCP client
claude · acme-shopuptimeeye mcp · connected
Alert · #alerts-prod · 14:05

Checkout flow is DOWN: step 3 POST /checkout/confirm returned 502 from eu-central-1 and us-east-1.

You

The checkout monitor went down at 14:05. What did the checkout service log around then, and is that error new?

Claude
query_logsservice:checkout-svc level:errorlogs_histogramget_log_context
14:04:58warnupstream latency high target=payments-gw p99=1840ms
14:05:02errorPOST /checkout/confirm 502 err="upstream timeout" trace=9f3ac2
14:05:03errorretry 1/3 failed status=502 trace=9f3ac2
It's new. 0 occurrences in the previous 24 h, 212 since 14:02 — the same minute payments-gw was deployed with UPSTREAM_TIMEOUT=2s (was 10s). Shall I raise the timeout and open a PR?
You

Yes, and tell me when the monitor is green.

eu-central-1 · us-east-1 · 30 s interval
Checkout flow recovered · 14:11
MCP server · included

One line. Then just describe it.

Same API, same permissions as the app, no server to run. 25 tools: monitors, scheduled tasks, notification channels, incidents, and five read-only logs tools your assistant uses to explain an outage.

Read the MCP docs
# Claude Code
claude mcp add --transport http uptimeeye https://api.uptimeeye.com/mcp \
  --header "Authorization: Bearer ue_live_XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX"
# then, in your session

Monitor https://shop.example.com/checkout as a 3-step flow (login → add to cart → confirm) every 30 s from EU and US, alert #alerts-prod when a step fails.

create_monitor · create_notification_channel

One incident, end to end

The deploy that broke checkout at 16:40.

Without monitoring you hear about it from the first support ticket on Monday. This is what happens instead. Every beat on the right is a feature on every plan, free included.

Everything in it is on every plan
  1. failing

    Step 3 of the checkout flow fails

    A single uptime check still says 200 OK. The multi-step monitor runs login → cart → confirm and sees the 502 on the one request that takes the money.

    Checkout flow · productioneu-central-1
    1. 1.POST /api/login89 ms
    2. 2.POST /api/cart/items142 ms
    3. 3.POST /api/checkout/confirm502
    4. 4.GET /api/orders/:id—

    Multi-step monitors · up to 10 steps

  2. confirmed

    A second location sees the same 502

    us-east-1 reports the same failing step. Not flaky, not a network blip: it's real. Only now does the alert fire, so you get outages, not a false-alarm storm.

    Verified from multiple locations · up to 6

  3. alerting

    Slack and PagerDuty get the exact failing step

    The step, the status code and the locations, with a link to the logs of that step. Nobody opens a dashboard to find out what broke.

    UptimeEye APP · #alerts-prod · just now
    Checkout flow is DOWNStep 3 (POST /checkout/confirm) failed in eu-central-1 and us-east-1 · 502
    View logs of this step · Acknowledge

    Slack, PagerDuty, Teams, Discord, Telegram, email, webhook

  4. on-call

    Ack from the phone, open the logs

    The alert links to the log lines of that step. No second tool, no copying a timestamp into another search box.

    service:checkout-svc16:40 – 16:45 · 4 matches
    16:40:12infoPOST /cart/items status=200 dur=142ms
    16:40:58warnupstream latency high target=payments-gw p99=1840ms
    16:41:02errorPOST /checkout/confirm 502 err="upstream timeout" trace=9f3ac2
    16:41:03errorretry 1/3 failed status=502 trace=9f3ac2
    Hand off to Claude Code
    checkout-svc · POST /checkout/confirm 502 err="upstream timeout" · 212× in the last 24 h
    Copy prompt

    A log alert on this error signature pages you before the next check.

    Logs included · 1 GB free · EU-hosted

  5. assistant

    Ask Claude. It reads the same logs over MCP.

    You

    The checkout monitor went down at 16:41. What did checkout-svc log around then, and is that error new?

    Claude
    query_logslogs_histogramget_log_context
    New since 16:40. 0 occurrences in the previous 24 h, 212 since. The timeouts start the minute payments-gw was deployed with UPSTREAM_TIMEOUT=2s (was 10s). I'd roll back or raise the timeout. Want a PR?

    MCP server · 25 tools · Claude Code, Cursor, VS Code

  6. recovered

    Rollback. Next check passes from all locations.

    The status page updates itself, the subscribers get the all-clear, and the incident closes with the full timeline attached.

    All systems operationalstatus.acme-shop.com

    Status pages · incidents · own domain on Pro

  7. done

    6 min 45 s. Zero tickets. Friday evening intact.

    ≈ 1 min
    Detection
    < 2 min
    First alert delivered
    6 min 45 s
    Total outage time
    0
    Support tickets

Stop losing customers to broken user flows

  • Catch checkout, login, and signup failures before your customers do - simulate complete user journeys in one monitor
  • Extract session tokens and data between steps - test like real users navigate your app
  • Spot breaking changes instantly with step-by-step assertions on status codes, response data, and headers
  • Debug faster with detailed execution logs showing exactly where and why flows break
  • Zero downtime. Zero frustrated customers.

Part of all plans - no setup needed

Login → Add to cart → Paid. Monitor what matters most.

Deliver consistent performance worldwide

  • Prevent regional outages from killing conversions - monitor critical flows from every continent
  • Spot slow endpoints before they impact user experience - track response times, DNS resolution, and timeouts globally
  • Get alerted the moment a step fails - every alert is verified from multiple regions first, so you get real outages, not a false-alarm storm
  • Detect SSL certificate problems and CDN failures before customers complain
  • Global reliability. Local peace of mind.

Part of all plans - no setup needed

Run checks from up to 6 regions (Agency; Pro 3, Free 1) to catch local slowdowns before your users do.

Turn silent failures into instant alerts

  • Catch failed cron jobs within seconds — not when customers complain hours later
  • Prevent data loss from missed backups, broken reports, and failed email campaigns
  • Transform mysterious failures into actionable insights with runtime error tracking and delay monitoring
  • Eliminate guesswork with precise failure notifications showing exactly what went wrong and when
  • Never miss a critical failure again.

Essential for production systems - included in all plans

Missed execution? Failed task? We'll tell you - instantly.

Fewer 'Is it down?' tickets with proactive communication

  • Stop the flood of 'Is it down?' tickets - keep stakeholders informed with real-time status updates
  • Share incident details automatically - uptime metrics, job failures, and performance issues in one place
  • Maintain trust during outages with branded status pages that match your company's look and feel
  • Reduce support workload while improving customer satisfaction through transparent communication
  • Proactive communication. Happier customers.

Professional status pages

Clear status pages that reduce support noise.

Your entire stack on one screen

  • See status, heartbeat history, uptime, and response time for every monitor in one live view
  • Spot slow responses and flapping services before they turn into outages
  • Filter and sort by status, tag, or response time to find trouble in seconds
  • Live counters for online monitors, incidents, and 24h averages - always current, no refresh needed
  • One look. Total clarity.

Part of all plans - no setup needed

All monitors, one glance - status, heartbeat, uptime, response.

From “it’s down” to “here’s why”, in one tab

  • Ship logs from every service over plain HTTP, Vector, Fluent Bit or OpenTelemetry: one ingest key, every JSON key a searchable field, no schema
  • Search with LogsQL, filter by service and level, and follow new lines as they arrive in live tail
  • Alert and logs in one place: when a step fails, the error lines of that minute are one filter away, and a log alert on the error signature pages you before the next run
  • Hand-off to Claude Code: one log line becomes a self-contained prompt with its context, frequency and related entries
  • 1 GB / 3 days on Free, 10 GB / 7 days on Pro, 50 GB / 15 days on Agency. Hosted in the EU.

Included in every plan

Alert → the error lines that caused it. Checkout flow, step 3, 14:05.

Built for developers.

Set it up by describing it, over the API, or as code.

MCP server for AI assistants

Connect Claude Code, Cursor or any other MCP client with an org API key and set up monitors, scheduled tasks and notification channels by describing them. Same API, same permissions, no server to run: most clients connect directly, Claude Desktop through the mcp-remote bridge, which needs Node.js.

Set up the MCP server

API-first by design, with an official Terraform provider.

Pricing plans

From side projects to agencies. Plans differ in volume and interval, never in features: every new workspace starts with 14 days of Pro, no credit card, Free forever afterwards.

€0/month

Forever free

Free

Real monitoring with real alerts — enough for any side project or small site. Commercial use explicitly allowed.

Start free
5 monitors (HTTP, DNS, Ping)
3 scheduled tasks
Multi-step flows (up to 5 steps)
1 status page
5-minute check interval
All alert channels — Slack, Discord, Telegram, MS Teams, PagerDuty, webhooks, email
Logs: 1 GB / month, 3-day retention
Commercial use allowed
Recommended
€29/month

Billed monthly, excl. VAT

Pro

Everything for a production service: 30-second checks, multi-region monitoring, your own domain on status pages, and your whole team on board.

Start free
50 monitors
50 scheduled tasks
30-second check interval
Checks from 3 regions
3 status pages
Custom domain for status pages
Up to 1,000 status-page subscribers
Up to 5 team members
Multi-step flows (up to 10 steps)
Logs: 10 GB / month, 7-day retention
All alert channels — Slack, Discord, Telegram, MS Teams, PagerDuty, webhooks, email
€79/month

Billed monthly, excl. VAT

Agency

For agencies & resellers: your own brand and domain on every status page, unlimited subscribers, and room for many client monitors.

Start freeMonitoring for agencies →
150 monitors
150 scheduled tasks
30-second check interval
Checks from 6 regions
10 status pages
White-label status pages — your logo & colors, no UptimeEye branding
Custom domains (up to 10)
Unlimited status-page subscribers
Up to 10 team members
Multi-step flows (up to 20 steps)
Logs: 50 GB / month, 15-day retention
Priority support

Frequently asked questions

Will this flood my Slack with useless alerts?

No. UptimeEye uses smart thresholds, location checks, and step-level assertions to only alert you when something really breaks - not when your site hiccups in one region.

How fast can I set this up?

You'll have your first monitor running in under 5 minutes - no scripts, no agents, no nonsense. Just point, click, and catch your first failure.

Can I monitor complex flows like login → cart → checkout?

Yes. UptimeEye lets you monitor full user journeys - login, signup, checkout - step by step. Pass tokens, check responses, catch silent failures before your users ever see them.

We already use Datadog, Hyperping or UptimeRobot. Why switch?

It depends on what you run there. At Datadog, 20 API tests every 60 seconds from 2 locations plus 10 GB of logs with 7 days retention come to about $890/month on an annual contract (list prices, 10 October 2026); UptimeEye Pro covers that setup for €29/month. Hyperping Essentials ($29/month) is comparable to Pro. UptimeRobot is a good fit for simple pings, but has no multi-step flows.

What if a cron job fails silently?

You'll know. Each job reports to UptimeEye when it runs; if a run is missing, late or fails, you get an alert - even if no one is watching.

Do you support single sign-on (SSO)?

Not yet. Today you sign in with your account and invite team members per organization (up to 5 seats on Pro, 10 on Agency). SAML and OIDC are on the roadmap — tell us if you need them and we'll prioritise by demand.

What if I need to convince my CTO/team?

Send them your live status page. Trigger a failure. UptimeEye picks it up in seconds - visible to everyone. It's the fastest way to prove value without a single meeting.

Is there documentation or an API?

Yes - and it's actually good. Full OpenAPI spec with org API keys, an MCP server for AI assistants, and an official Terraform provider — all included on every plan, plus setup guides and examples for CI/CD and Postman. Start here: https://uptimeeye.com/docs

We're an agency. Is there a plan for client sites?

Yes. The Agency plan (€79/month) covers 150 monitors and white-label status pages on your clients' domains. Details: https://uptimeeye.com/agencies

Give your assistant eyes on production.

Monitors, logs and the MCP server are on the free plan. Paste a cURL or describe your first flow to your assistant.

Start free

14 days of Pro included, then free forever. No credit card.