API Status Check MCP Server

Let Claude, Cursor or any MCP client answer “is this API down right now?” with a real probe instead of a guess. Hosted, public, no API key, nothing to install.

Endpoint (streamable HTTP, POST)
https://apistatuscheck.com/api/mcp

Why an agent wants this

A model asked whether a provider is down has three bad options: answer from training data that predates the outage, fetch a status page written for humans and hedge, or say it cannot know. This server gives it a fourth — a call that returns up, degraded or down with a response time and a timestamp, across 285 APIs in 30 categories.

Setup

Claude Code

claude mcp add --transport http apistatuscheck https://apistatuscheck.com/api/mcp

Claude Desktop, Cursor, Windsurf and other config-file clients

{
  "mcpServers": {
    "apistatuscheck": {
      "type": "http",
      "url": "https://apistatuscheck.com/api/mcp"
    }
  }
}

Verify it from a terminal

curl -s -X POST https://apistatuscheck.com/api/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Tools

get_api_statusCheck an API right now

Probes the API endpoint at call time and returns up/degraded/down with a response time, alongside the last hourly monitor reading so an agent can see the two disagree instead of reporting a blip as an outage.

Arguments: slug
search_apisSearch monitored APIs

Resolves a product name a user mentioned ("OpenAI", "payments") into the slug the other tools take. Matches on name, slug, category and description.

Arguments: query, limit?
list_apisList monitored APIs

The monitored catalog, optionally filtered by category. Statuses come from the most recent hourly run and carry a timestamp.

Arguments: category?, limit?
list_down_apisList APIs currently down or degraded

Everything the latest monitor run found unhealthy, across all categories. Returns monitorLastRun so an empty list can be told apart from a stale snapshot.

Arguments: —
get_uptime_historyGet retained uptime history

Uptime percentage, response-time stats and raw data points over 24h, 7d or 30d. History is retained 30 days; a recently added API reads as null rather than a misleading 100%.

Arguments: slug, range?
list_categoriesList API categories

The 30 category slugs and how many monitored APIs sit in each.

Arguments: —

Example response

get_api_status for openai. Note that errorMessage can be set on a healthy result: an authenticated endpoint answering 401 is proof it is reachable, which is the point of the probe.

{
  "slug": "openai",
  "name": "OpenAI",
  "live": {
    "status": "up",
    "responseTimeMs": 89,
    "checkedAt": "2026-08-01T18:34:26.261Z",
    "errorMessage": "HTTP 401 (reachable)"
  },
  "monitor": {
    "status": "up",
    "description": "All Systems Operational",
    "lastChecked": "2026-08-01T14:07:52.457Z"
  },
  "officialStatusPage": "https://status.openai.com",
  "url": "https://apistatuscheck.com/openai"
}

Limits worth knowing before you rely on it

  • One region, one probe. get_api_status is a single request from a single place. It is evidence, not consensus — compare it against the monitor field in the same response before calling an outage.
  • An empty list is not an all-clear. list_down_apis returns nothing when the monitor snapshot is missing or stale as well as when everything is healthy. Check monitorLastRun.
  • History is 30 days. Recently added APIs return null uptime percentages rather than a flattering 100%.
  • POST only. The transport is stateless streamable HTTP with JSON responses. GET and DELETE return 405; SSE-only clients will not connect.

FAQ

Do I need an API key to use the API Status Check MCP server?

No. The endpoint at https://apistatuscheck.com/api/mcp is public and unauthenticated. Add the URL to your MCP client and the six tools are available immediately.

Is there anything to install?

No. This is a hosted remote MCP server over the streamable-HTTP transport, not an npm package. There is no local process, no Node version to match and nothing to keep updated — the tool list your agent sees is always the deployed one.

How is get_api_status different from the hourly monitor?

get_api_status makes a real outbound request when your agent calls it, so it reflects the moment of the question. list_apis and list_down_apis read the most recent hourly monitor run and carry a timestamp. get_api_status returns both values side by side, because a single probe from one region is weak evidence on its own.

Which APIs are covered?

The catalog is 285 APIs and developer platforms across 30 categories — AI providers, cloud, payments, communication, developer tools, databases, email and more. Call list_categories or search_apis to see what is monitored.

Does it support the SSE transport?

No. The endpoint is POST-only streamable HTTP with JSON responses; it is stateless so that it stays correct across serverless instances. GET and DELETE return a 405. Clients that require the older SSE transport will not connect.

Prefer plain REST? The same data is available without MCP — see the JSON API docs and the status badges.