Crontinel
Cron, queue and agent monitoring
About Crontinel
Crontinel is an open-source monitoring tool for the background work that keeps an app alive but never shows up on an uptime dashboard: cron schedules, queue workers, supervisors, backups and AI agent runs. The problem it exists to solve is a familiar and dangerous blind spot. A URL responding 200 tells you the web server is up; it says nothing about whether last night's scheduled job actually ran, whether a queue worker is draining or silently dead, or whether an AI agent is stuck in a 40-minute retry loop quietly burning tokens. Traditional uptime checks and status pages only confirm that an endpoint answers, so a cron entry a deploy clobbered, a stuck overlap lock, or a backlogged invoice queue all read as green while real work stops. Crontinel hooks directly into the scheduler and queue and alerts on the one signal that matters: the run that should have happened and did not.
What makes it distinct is that it is open-core and framework-native rather than a generic ping service. It ships MIT-licensed SDKs for Laravel, Node, Python, Go, Rust, PHP, Ruby, Rails and .NET, plus a language-agnostic CLI that can wrap any cron, Docker or Kubernetes job, and each package works standalone against your own database with no SaaS required. Installation is two commands and hooks into your framework's events instead of wrapping every task by hand, so there is no per-job instrumentation to maintain. It also treats AI agents as first-class monitored workers, tracking runs, tool calls, token spend, model latency and loop detection, and exposes an MCP server so your own agent can query "are any queues or agents backing up?" and get a real answer. The cloud dashboard is an option layered on top, not a dependency.
You install one SDK for your runtime, which auto-detects your scheduler and queue driver, then run an init command and, optionally, add crontinel check to your CI or deploy hook. From that point Crontinel reads scheduler events, worker state, queue internals and agent traces directly, without endpoint pings or per-task wrappers. Every scheduled task registers a run attempt, exit code and duration, so a missed run trips an alert before the next cron tick, typically within 60 seconds. Workers report per-supervisor status, queue depth and failed-job rate; agents report runs, tool calls, token spend and loop detection. When something drifts, Crontinel routes a single alert to the channels your team actually watches - Slack, PagerDuty, email or an HMAC-signed webhook - with per-service routing, quiet hours and auto-resolve so one problem does not fan out into noise. The dashboard shows the inside of your queue rather than a heartbeat chart: every task with its last run and exit code, every worker with its state, every queue with depth and oldest job, and every agent run with its tool calls. You can self-host the whole thing or use the hosted cloud for multi-app views and longer history.
- •Cron And Heartbeat Monitoring - Every scheduled task registers a run attempt, exit code and duration, so a missed or late run is detected in under 60 seconds instead of surfacing hours later from an angry user.
- •Queue And Worker Visibility - Per-worker status, queue depth trends, oldest-job age and failed-job rate expose a stalled or backlogged queue that a "running" process count would hide.
- •AI Agent Monitoring - Agents are watched like any other worker, tracking runs, tool calls, token spend, model latency and loop detection so a runaway agent is caught before it shows up on the bill.
- •Framework-Native SDKs - MIT-licensed packages for Laravel, Node, Python, Go, Rust, PHP, Ruby, Rails and .NET hook into framework events with no per-task wrapping, plus a CLI that wraps any cron, Docker or Kubernetes job.
- •Routed Alerting - One alert fans out to Slack, PagerDuty, email or HMAC-signed webhooks with per-service routing, quiet hours and auto-resolve, so the right failures reach the team without drowning it in noise.
- •MCP Server For Agents - An MCP server lets your own agent query jobs, queues and runs directly, so it can answer questions about backlogs and stalls with real data rather than guesses.
Crontinel is built for backend and platform engineers who run scheduled work in production and have been burned by a silent failure that no uptime tool caught. It fits Laravel and Horizon shops especially well, since it began as the actively maintained open-source alternative in that ecosystem, but its multi-language SDKs and CLI make it equally useful for Node, Python, Go and Rails teams, or anyone wrapping Docker and Kubernetes jobs. Because every package is MIT-licensed and works standalone, it suits solo developers and open-source-minded teams who want full cron and queue visibility on their own infrastructure with no SaaS dependency, as well as larger teams that want the hosted cloud for multi-app dashboards, role-based access and long-horizon history. It is also a natural fit for teams running AI agents in production who need to treat agent runs, token spend and loop detection as monitored infrastructure rather than an unobserved cost center.
