Initialize the MCP server
AlertsDown connects to the endpoint with the selected transport and verifies that the server can complete the MCP initialization flow.
MCP Server Monitoring
AlertsDown can monitor MCP endpoints your agents depend on. Add your own or vendor MCP server, define the tools, resources, prompts, and safe probe that must keep working, then receive alerts through your existing downtime notification channels when the MCP contract changes or fails.
The difference
The AlertsDown hosted MCP page is for connecting AI clients to AlertsDown. Your assistant uses that endpoint to create monitors, inspect status, and work with uptime data.
This page is the other direction: AlertsDown connects to MCP servers you add. It verifies the server still speaks MCP, still advertises the capabilities your agents need, and can still answer one safe read-only tool call if you choose to configure a probe.
What the monitor checks
AlertsDown connects to the endpoint with the selected transport and verifies that the server can complete the MCP initialization flow.
The monitor reads the tools, resources, and prompts the server advertises so missing capabilities are caught before agents rely on them.
Add exact tool, resource, or prompt names. AlertsDown alerts when a deploy removes or renames something your agents need.
Configure one read-only tool call, such as ping, health, or list_status, to verify that the tool is still callable end to end.
How MCP alerts work
An MCP monitor stores the endpoint, transport, authentication, expected capabilities, optional probe details, and notification channels. On each scheduled check, AlertsDown validates the MCP server from outside your app. Failed initialization, missing expected capabilities, and failed probe calls are treated as monitor failures and routed to the contacts you selected.
Setup in AlertsDown
MCP Server Monitoring is available on Indie Builders and above. Use it for your own MCP services, internal agent tools, or vendor endpoints where uptime alone is not enough.
Sign in, open My URLs, choose Add URL, then switch Monitor Type from HTTP / Website to MCP Server.
Paste the Streamable HTTP or SSE endpoint agents use, choose the transport type, timeout, and check interval.
Select no auth, bearer token, or custom header depending on how your MCP server protects the endpoint.
List important tools, resources, and prompts by exact name, then enable alerts for missing capabilities.
Optionally call one non-destructive read-only tool with JSON arguments and alert if the probe fails.
Route failures through your AlertsDown contacts, including email, SMS, Telegram, WhatsApp, or webhook channels available on your plan.
Name: Production MCP Server. Endpoint: https://app.example.com/api/mcp.
Transport: Streamable HTTP. Auth: bearer token or custom header. Expected tools:
search_docs, create_ticket, list_incidents. Probe:
health_check with {"verbose":false}.
Use cases
Get alerted when the MCP server behind your internal coding, search, database, or deployment tools stops initializing correctly.
Catch deploys that rename or remove a required MCP tool, resource, or prompt before your agents start failing user tasks.
Track third-party MCP endpoints your team depends on and route failures to the same notification channels as website outages.
After deploys, migrations, auth changes, or dependency updates, verify that the MCP endpoint and safe probe still work.
Monitor MCP wrappers around databases, docs, ticketing, deployment systems, analytics, and incident tools.
Turn MCP monitor failures, recovery timing, and probe details into postmortem notes, customer updates, and follow-up tasks.
How to use it
Monitor the production MCP endpoint and alert if initialization fails.
Alert if required tools likesearch_docs,list_incidents, orcreate_ticketdisappear.
Add a safe probe for health_check and alert if it fails after deployment.
Send MCP monitor failures to the same on-call contacts as API and website downtime.
Access and security
Use bearer tokens or custom headers for private MCP endpoints. Store only the secret needed for the monitor to initialize and run safe checks.
Probe calls should be read-only and non-destructive. Use health, list, status, or ping-style tools rather than write or mutation tools.
MCP Server Monitoring is available on Indie Builders and above, separate from the hosted AlertsDown MCP setup for AI clients.
Add your endpoint, pin the capabilities that matter, choose a safe probe, and route failures to your AlertsDown alert channels.
MCP alerts are AlertsDown monitors for your own or vendor Model Context Protocol servers. They notify you when an MCP endpoint stops initializing, loses expected capabilities, or fails a configured safe probe.
MCP alerts check MCP servers you add to AlertsDown. Hosted MCP is AlertsDowns own MCP endpoint that AI tools connect to when they need to manage your AlertsDown monitors.
It checks endpoint connectivity, MCP initialization, advertised tools, resources, prompts, expected capability names, and optional safe read-only probe calls.
MCP Server Monitoring supports Streamable HTTP and SSE transports in the AlertsDown monitor setup flow.
MCP Server Monitoring is available on the Indie Builders plan and above.