# Capto > A dead-man's switch monitoring tool for cron jobs, workers, and containers that alerts you when a job goes quiet and shows the last breadcrumb context it left behind. Capto is a background job monitoring service that acts as a dead-man's switch for cron jobs, workers, and containers. When a monitored job stops pinging, Capto fires an alert and surfaces the last structured breadcrumbs the job left behind — so you get context alongside the notification, not a dashboard to dig through. It requires no agent installation and no open ports. ## What It Is Capto watches background jobs through three check types: heartbeat (a worker pings a URL on each healthy run), cron (a schedule and timezone define when a ping is expected), and HTTP (Capto polls an endpoint and requires consecutive failures before alerting). The core design principle is that the alert arrives with the last useful context — structured breadcrumb lines the job posted while it was alive — frozen onto the incident at the moment the check goes down. ## How the Breadcrumb Model Works While a job is running, it can post short structured log lines ("breadcrumbs") to Capto via a simple API call. These are held in a ring buffer and cost nothing until something breaks. When a check goes down, the most recent breadcrumbs are snapshotted onto the incident and travel with the alert to ntfy, Telegram, email, or a webhook. A `/fail` ping can also carry the job's full output (up to 10 kB), so the traceback arrives with the page rather than requiring a separate lookup. ## Detection and Reliability Guarantees Capto makes several explicit reliability commitments: - **Detection is independent of notification** — muting or quiet hours suppress the alert, never the evaluation. A silenced check that goes down still reads as down. - **Alerts are queued before sending** — each notification is a durable row per channel, retried with exponential backoff. A downstream outage (e.g., ntfy or Telegram) delays the alert rather than dropping it. - **One open incident per check** — enforced by a unique database index, not a read-then-write, so repeated failures cannot open duplicates or go silent after the first alert. - **Cron checks understand schedules** — the last ping is compared against the previous scheduled fire, so weekly and monthly jobs behave correctly. Timezone is validated at save time, not at 3am. - Every check is evaluated every 30 seconds on every plan, because the evaluator runs a single loop over all active checks. ## Integration and Setup Path The minimal integration is one `curl` call at the end of a job. Capto also supports `/start` and `/fail` URL suffixes to distinguish jobs that hung from jobs that never started. For fleets of short-lived workers, an ingest key allows checks to self-provision on the first ping — no dashboard setup required. Docker, Podman, and systemd recipes use a sidecar or host cron to ping while the container is healthy; Capto never needs access to the Docker socket. A TypeScript client snippet is provided in the docs, with an npm package noted as forthcoming. ## What It Deliberately Is Not The About page explicitly declines several common monitoring features on purpose: Capto is not a log platform (breadcrumbs are a short ring buffer, not a searchable history), not an APM or metrics product (no traces, no dashboards, no query language), and not an on-call scheduling tool (no rotations or escalation policies). The stated design philosophy is that scope is a feature. ## Current Status Capto is actively available with a free tier (20 checks, no credit card required) and a paid Pro plan. The product blog documents ongoing decisions about detection, pricing philosophy, and feature scope. The About page includes a shutdown commitment: at least 90 days' notice by email, API access maintained through that window, and no billing past the announcement date. ## Features - Heartbeat monitoring for workers and daemons - Cron job monitoring with schedule and timezone support - HTTP endpoint polling with consecutive-failure threshold - Breadcrumb context snapshotted onto incidents at alert time - /start and /fail ping URL suffixes for run lifecycle tracking - Auto-provisioning checks via ingest key for worker fleets - Durable alert outbox with exponential backoff retry - One open incident per check enforced by unique index - Mute, quiet hours, and pause controls with independent evaluation - Alerts via ntfy, Telegram, email, and webhook - Detection every 30 seconds on all plans - No agent to install, no port to open - Docker, Podman, and systemd heartbeat recipes - TypeScript client snippet included in docs - Export all data via API with your own token ## Integrations ntfy, Telegram, Email, Webhook, Docker, Podman, systemd, curl, Python requests ## Platforms WEB, API, CLI ## Pricing Freemium — Free tier available with paid upgrades ## Links - Website: https://capto.run - Documentation: https://capto.run/docs - EveryDev.ai: https://www.everydev.ai/tools/capto