MCP Integration

Updated April 15, 2026

The Model Context Protocol (MCP) server exposes Hyperping's monitoring data and controls to AI agents over HTTP. Works with any MCP-compatible client: Claude Desktop, Claude Code, Cursor, Windsurf, mcp-inspector, or custom clients built on the MCP SDK.

📘Machine-readable spec

For a dense, agent-friendly spec (tools, query patterns, response shapes), point your client at /mcp/llms.txt. Zero-setup context.

Endpoint

URL

https://api.hyperping.io/v1/mcp

Methods

POSTGETDEL

Transport

Streamable HTTP · JSON-RPC 2.0

Auth

Bearer token

Scope

One project per key

Tools

25 read · 5 write

Quick start

Step 01

Create an API key

Go to Project Settings → API Keys and create a read_only or read_write key. Keys are strictly scoped to one project.

Step 02

Add to your client

Paste the mcpServers block below into your Claude Desktop, Cursor, or Windsurf config.

Step 03

Restart and ask

Restart your client. Start asking questions: "What's down?", "SLA on the API monitor this month?"

Client configuration

Claude Desktop, Cursor, Windsurf, Claude Code. Add to the mcpServers block of your client config:

Streamable HTTP (recommended)JSON

{
  "mcpServers": {
    "hyperping": {
      "type": "http",
      "url": "https://api.hyperping.io/v1/mcp",
      "headers": {
        "Authorization": "Bearer sk_..."
      }
    }
  }
}

Older clients that only speak stdio can bridge via mcp-remote:

stdio bridge via mcp-remoteJSON

{
  "mcpServers": {
    "hyperping": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://api.hyperping.io/v1/mcp",
        "--header",
        "Authorization: Bearer sk_..."
      ]
    }
  }
}

Tool reference

Your AI composes the queries; Hyperping returns the facts. Writes are scoped to reversible operations, delete_monitor is intentionally not exposed.

Monitors 6 tools

list_monitors(status?, page?, limit?)read

Paginated list with optional filter: up, down, paused, ssl_expiring.

get_monitor(uuid)read

Single monitor with full config.

search_monitors_by_name(query)read

Case-insensitive substring match across names and URLs.

create_monitor(...)write

HTTP, ICMP, port, or DNS. Accepts regions, frequency (10s to 24h), expected status, keyword, headers, escalation policy.

update_monitor(uuid, ...)write

Partial update.

pause_monitor(uuid) / resume_monitor(uuid)write

Toggle without losing history.

Status & reporting 5 tools

get_status_summary()read

Counts plus list of down monitors with duration.

get_monitor_uptime(from?, to?, resolution?, monitor_uuids?)read

SLA %, aggregate plus per-day, hour, week, or month.

get_monitor_response_time(...)read

Latency trend.

get_monitor_mttr(...)read

Mean time to resolve, per monitor.

get_monitor_mtta(...)read

Mean time to acknowledge, per monitor.

Outages 4 tools

list_outages(page?, search?, status?, type?)read

Paginated.

get_outage(uuid)read

Flat record.

get_outage_timeline(uuid)read

Detection, cross-region verification, alerts dispatched, acknowledgement, resolution.

get_monitor_outages(monitor_uuid, page?, status?)read

Scoped to one monitor.

Observability 3 tools

get_monitor_anomalies(uuid)read

Anomaly detection (flapping, latency spikes).

get_monitor_http_logs(uuid, page?, limit?, level?)read

Recent probe logs.

list_recent_alerts(from?, to?, resolution?, monitor_uuids?)read

Notification history across all channels.

On-call context 5 tools

list_on_call_schedules()read

All current rotations.

get_on_call_schedule(uuid)read

Rotation detail with overrides.

list_escalation_policies()read

Full policy chain.

get_escalation_policy(uuid)read

Policy detail.

list_team_members()read

Name and email for user-ID lookups.

Integrations 2 tools

list_integrations()read

All configured channels: Slack, Telegram, Discord, PagerDuty, OpsGenie, Teams, webhook, email.

get_integration(uuid)read

Channel-specific config.

Security

Project scope

API keys are bound to exactly one project in the database (FK-constrained). req.projectuuid is resolved server-side from the key. Client headers cannot override.

Row-level isolation

Every controller SQL scopes by projectuuid = ?. Cross-project access returns 404, never leaks.

Read-only enforcement

Write tools check apiKeyPermissions === 'read_only' and refuse with a clear error. Enforced at the tool handler, not at the route.

Role gating

Session-auth endpoints (reporting, schedules, on-call, integrations) run hasAccountRole with the user's actual project role. viewer keys cannot invoke write-adjacent endpoints.

Revocation

Deactivating a key in the dashboard blocks the next MCP call immediately. No TTL to wait out.

Not available via MCP

⚠Scoped out by design

delete_monitor. Pause instead, or delete via dashboard / REST API.
Billing, API key management, SSO, teammate invites, password reset.
Status-page incidents (/v3/incidents/*). Different resource. Contact us if you need them exposed.

Limits

Pagination

Up to 200 monitors or outages per page.

Rate limit

Shared with the REST API: 300 req / min / key.

Response size

Large windows (for example, 30-day alert history) can be capped by narrowing from / to.

Next steps

REST API

API Reference →

Direct HTTP access to the same resources. Same auth, same rate limits, same activity log.

SDKs

Community libraries →

Terraform provider, Python SDK, Go client, Prometheus exporter. All talk to the same REST API under the hood.

Quotas

API rate limits →

Exact caps and how to raise them for enterprise agents.

Support

Missing a tool? Want status-page incidents exposed? Tell us what your agent needs.

← Docs home REST API reference →