Beta

The MCP server that gives your AI agent eyes on your infrastructure

The Xitoring MCP server exposes your servers, uptime checks, incidents, and metrics to any MCP-compatible client. Ask questions in plain language, and let your agent act — 70 tools over one HTTP endpoint.

Get your API key Read the docs

terminal

claude mcp add --transport http --scope user xitoring \
  https://app.xitoring.com/mcp \
  --header "Authorization: Bearer YOUR_API_KEY"

Social Proof

Trusted by thousands — rated on

See what real users say about Xitoring on the world's top review platforms.

What is the Xitoring MCP server?

The Model Context Protocol (MCP) is an open standard that lets AI assistants talk to external systems through a consistent interface. Instead of pasting dashboard screenshots into a chat window, your assistant queries the source directly and reasons over live data.

The Xitoring MCP server is a hosted HTTP endpoint that puts your entire monitoring account behind that interface. Point any MCP client at https://app.xitoring.com/mcp with your API key and it gains 70 tools spanning live metrics, uptime and SSL checks, incidents, status pages, reports, and machine-learning insight — plus a curated set of write actions for resolving incidents and managing checks.

Setup

Connect in 60 seconds

One HTTP endpoint, one bearer token. No agent to install, no gateway to run.

Generate an API key

In the Xitoring dashboard, go to Account → API Access and create a key.

Add the server

Start asking

Your agent discovers all 70 tools automatically. No further configuration.

terminal

claude mcp add --transport http --scope user xitoring \
  https://app.xitoring.com/mcp \
  --header "Authorization: Bearer YOUR_API_KEY"

Run this once. The --scope user flag makes the server available in every project.

The MCP server is in beta. The endpoint and authentication are stable; individual tools may still change.

Capabilities

70 tools, 8 resources

Everything your agent can read and do. 52 tools are strictly read-only; 18 can change your account.

52 Read-only18 Write

Monitoring

Read-only16 tools

Live and historical metrics straight off your hosts — CPU, memory, disk, network, running processes, and fleet-wide rankings.

cpu_load_average
cpu_usage
memory_usage
disk_usage
disk_io
network
check_uptime
ongoing_incidents
current_maintenance_schedules
upcoming_maintenance_schedules
top_processes_by_cpu
top_processes_by_memory
server_status
top_servers_by_cpu
top_servers_by_memory
top_servers_by_disk_io

Resource

Read-only16 tools

List and inspect the objects in your account: servers, checks, incidents, triggers, status pages, maintenance windows, and notification roles.

list_servers
get_server
list_checks
get_check
list_incidents
get_incident
list_triggers
get_trigger
list_status_pages
get_status_page
list_maintenance_schedules
get_maintenance_schedule
list_notification_roles
list_ssl_checks
get_trigger_options
recommend_triggers

Extended Monitoring

Read-only3 tools

Aggregate performance figures — response time, overall uptime, and average latency across a period.

get_check_response_time
get_overall_uptime
get_average_latency

Extended Resource

Read-only10 tools

The long tail: heartbeat checks, incident notes, contacts, subscription, tickets, uptime and SSL reports, notification logs, Xitogent logs, and account audit logs.

list_heartbeat_checks
get_incident_notes
list_contacts
get_subscription
list_tickets
get_uptime_report
get_notification_logs
get_xitogent_logs
get_ssl_uptime_report
get_account_logs

Actions

Write18 tools

The only tools that change state. Resolve incidents, pause and unpause checks and servers, create HTTP, ping, SSL, and heartbeat checks, schedule maintenance, open tickets, and manage triggers.

send_test_notification
add_incident_note
create_ticket
pause_check
unpause_check
pause_server
unpause_server
resolve_incident
create_http_check
create_ping_check
create_ssl_check
update_check
test_check
create_maintenance_schedule
add_heartbeat_check
invite_contact
create_metric_trigger
update_metric_triggers

Write tools are the only ones that mutate your account. Every other tool is read-only.

ML Insight

Read-only7 tools

Machine-learned analysis over your history: anomalies, capacity forecasts, health summaries, risk scores, root cause analysis, and incident digests.

get_anomalies
get_capacity_forecast
get_server_health_summary
get_risk_score
get_root_cause_analysis
get_group_root_cause_analysis
get_incident_digest

ML tools are observe-only. They report what the models see and never raise an incident on their own.

How anomaly detection works

MCP resources

8 resources

Read-once, reuse-often context your agent can pull in without spending a tool call.

xitoring://docs/concepts
xitoring://docs/trigger-patterns
xitoring://account/profile
xitoring://notification-roles
xitoring://incidents/active
xitoring://server/{id}
xitoring://check/{id}
xitoring://incident/{id}

Full reference in the docs

70 tools · 8 resources

In practice

Ask your agent, not your dashboard

Real questions the MCP server can answer against your live account.

Rank my infrastructure by CPU usage right now.

Pulls top_servers_by_cpu across your fleet, then drills into top_processes_by_cpu on the worst offender to name the process responsible.

Draft a post-mortem from last night's incident.

Reads the incident timeline, notes, and notification logs, correlates them against server metrics from the same window, and writes the summary.

Which checks have been flapping this week?

Scans uptime history and response times across your checks, flags the unstable ones, and suggests trigger changes to quiet the noise.

Schedule maintenance for tomorrow 02:00–04:00 on the database group.

Creates the maintenance window, attaches the right servers, and confirms which alerts will be suppressed.

Are we going to run out of disk anywhere this quarter?

Runs the capacity forecast across every host and lists the ones projected to hit their threshold, with the date each is expected to cross it.

Why did the checkout API slow down at 14:00?

Runs root cause analysis over the affected window, correlating response times with host metrics and recent incidents to point at the likely cause.

Access you stay in control of

Bearer-token auth

The endpoint authenticates every request against your Xitoring API key. No OAuth dance, no separate credential store, no long-lived session.

Treat keys as passwords

An API key carries full account access. Store it in a secrets manager, keep it out of source control, and rotate it every 90 days.

Read-only by default

52 of the 70 tools cannot change anything. The 18 write actions are clearly separated, and revoking the key cuts off access instantly.

MCP server questions

What teams ask before wiring Xitoring into their AI tooling.

What is an MCP server, in plain terms?

MCP — the Model Context Protocol — is an open standard for connecting AI assistants to external systems. An MCP server is the adapter on the system side. The Xitoring MCP server translates your monitoring account into tools an assistant can call, so it works with real data instead of whatever you paste into the chat.

Which clients are supported?

Any client that speaks streamable HTTP MCP. We document Claude Code, Claude Desktop, Cursor, and VS Code, and the same endpoint and Authorization header work with any other compliant client.

Do I need to install or host anything?

No. The server is hosted at https://app.xitoring.com/mcp. There is no local process, no gateway, and no extra agent — Xitogent is only needed for server metrics collection, exactly as it is without MCP.

It says beta. What does that mean?

The endpoint and authentication are stable and safe to build on. Individual tool names, parameters, and response shapes may still change as we refine them, so pin nothing and expect the occasional improvement.

Can my agent change my monitoring setup?

Only through the 18 write tools, which are listed explicitly in the Actions group — things like resolving an incident, pausing a check, or creating a maintenance window. The other 52 tools cannot mutate anything. Most MCP clients also ask you to approve a tool call before it runs.

Does it cost extra?

No. The MCP server is included with your Xitoring account at no additional charge, on every plan.

What data leaves my account?

Only what your agent requests, and it goes only to the AI client you configured. Xitoring does not send your data anywhere else and does not train models on it. Bear in mind that whatever your assistant reads becomes part of that assistant's conversation context.

How do I revoke access?

Delete the API key in Account → API Access. The MCP connection stops working immediately — no separate revocation step, and no need to touch the client config.

Ready to stop guessing?

Start monitoring your infrastructure in under 60 seconds. No credit card required.

Start Free Trial Talk to Sales

Free plan available • No credit card required • Setup in 60 seconds

Keep exploring

More from Xitoring

AIOps Anomaly Detection Server Monitoring Uptime Monitoring API Monitoring Xitogent Agent