HookPulse

A dead man's switch for cron jobs and webhooks: your job pings a URL; silence beyond the interval becomes an alert.

Live: https://hookpulse.net · API index: https://hookpulse.net/api/ · OpenAPI: https://hookpulse.net/openapi.json · MCP server: https://hookpulse.net/mcp · Docs for agents: https://hookpulse.net/llms.txt

What it does

HookPulse watches for silence. A backup that did not run, a cron that died, a webhook that stopped arriving — none of them raise an error, so HookPulse turns the absence into one.

Create a monitor (name, interval), get a private check-in link, and make the job call it: curl -fsS https://hookpulse.net/in/ID at the end of the script, or a GET/POST from n8n, Make, Zapier or a GitHub Action. Every five minutes HookPulse compares the last ping with the interval; a miss goes out by e-mail or as a JSON POST (type: hookpulse.miss, last_event_at, interval_sec) to your alert URL, at most once every 24 hours per monitor. The last pings, latency and a body preview stay on the monitor page; a status page is available as JSON and RSS.

Quickstart (no account)

Snippets for curl, cron and n8n, and the exact JSON posted on a miss:

curl -s https://hookpulse.net/api/templates

Self-described API index (how to create a monitor without an account):

curl -s https://hookpulse.net/api/

What your job runs at the end (replace YOUR_ID):

curl -fsS https://hookpulse.net/in/YOUR_ID

MCP server (remote, nothing to install)

Streamable HTTP over the same public API. Every tool is a call on this API; operationId in the OpenAPI is the tool name. Server card: GET https://hookpulse.net/mcp.

Claude Code:

claude mcp add --transport http hookpulse https://hookpulse.net/mcp

Cursor (.cursor/mcp.json):

{
  "mcpServers": {
    "hookpulse": {
      "url": "https://hookpulse.net/mcp"
    }
  }
}

Main tools: create_endpoint, list_endpoints, get_endpoint, list_events, ping_ingest, status_feed_url, list_templates, create_guest.

Pricing

Free. Ten monitors, intervals from 90 seconds, one alert e-mail, unlimited pings and a status page, without an account. Confirm an e-mail and get 90 days without limits.

Paid. Per request (x402) or prepaid credit: an extra monitor US$ 0.10, intervals under 90 seconds US$ 0.05, an extra alert e-mail US$ 0.10.

What it is not

It is not uptime monitoring: it never calls your endpoints; your job calls it.

Files in this repository

README.md	this page
openapi.json	OpenAPI 3 document; operationId = MCP tool name
llms.txt	short guide for agents (endpoints, auth, pricing, MCP)
llms-full.txt	full reference: params, body, response field by field, errors (when published)
apis.json	APIs.json index (APIs.io)
mcp/server.json	manifest published in the Official MCP Registry (net.hookpulse/hookpulse)
okf/	OKF bundle: markdown with front-matter (index, about, API, FAQ)
examples/	curl.sh and python.py with the free calls above

About this repository

This repository mirrors the public, machine-readable surfaces of HookPulse as served at https://hookpulse.net: the OpenAPI document, the llms.txt guide, the MCP registry manifest, the OKF bundle and the APIs.json index. The product itself is not open source; the mirror exists so the API and the MCP server can be found, read and linked from here. Files are regenerated from the live surfaces (last sync: 2026-09-05); if something here disagrees with the site, the site is right.

Issues and suggestions are welcome here. Contact: contato@hookpulse.net. Author: Wendel (@wrpbr).