Know if it’ll make it.

Switchyard watches the scheduled jobs your launches depend on. When one runs late, you see which date is at stake, how much room is left, and what happens if the fix takes another day.

Start monitoring free Rehearse a slip

Free for 20 checks. No card needed. Your first check takes about a minute.

Delay the train

Illustrative

Warehouse loadjob · pings every night
Prepare reportwaits on the load
Customer export 09:00terminus · due 09:00

Delay the warehouse load by 0 min

Export ready08:40

Absorbed by slack0 min

Past the deadline0 min

Illustrative only: 20 minutes of slack before a 09:00 deadline. Real rehearsals use your own jobs, tasks and dates, and never change a check or an alert.

A late job is a date at risk

Next stop Live wall

Jobs ping Switchyard when they run, so their state is observed, not typed in. Tasks and launches wait on those jobs. When the warehouse load stalls, the customer export it feeds shows up as a deadline in danger, with the latest time the load can be fixed and still make it.

How to read the map

A line is a launch. Its tasks are the stops; the date it must ship by is the terminus.
A job feeding a task is a transfer. Its trains are pings, seen as they pass.
Dashes are planned service. Where the chain is projected to land, never what was observed.
A red diamond is a signal failure. A job observed down, and the arrivals it holds up.

A wall for the spare screen

Next stop Status page

Every job is a row on the board that ticks over when it pings, and flips to late, then down, the moment it falls behind. It stays connected and updates on its own.

Departures 4 checks

Queue consumer every 30 seconds

Invoice sync every 5 minutes

Search index rebuild every hour

Cache warmer every minute, staging

Ping, run finished
Run started
Run failed
Down: no pings, track closed

Illustrative simulation, sped up so a few minutes pass in half a minute. Invoice sync misses its schedule: Switchyard marks it late, then down once the grace period runs out, and sends an alert. When it pings again, you get a recovery alert.

A status page your customers can read

Next stop Slow runs

Choose which jobs are public and share one address. Anyone can see current status and 90 days of daily uptime, without an account.

Service status /status/acme

All systems running

Queue consumer99.93%

Invoice sync99.96%

Nightly database backup100.00%

90 days agoToday

Slow runs, before they become missed runs

Next stop Alerts

Ping /start when a job begins and Switchyard times every run. After eight runs it knows what normal looks like for that job, and flags a run that takes at least one and a half times its median. Those typical run times are also what the deadline projections use.

median 4 min

Search index rebuild ran slow 9 min 6 s. Usually 4 min.

Alerts where your team already looks

Next stop Setup

Slack, Discord, or your own endpoint, signed with HMAC-SHA256. One message when a job goes down, one when it recovers, one for a slow run. A down alert says which dates it blocks and the time it must be fixed by.

POST your endpoint

X-Switchyard-Signature: sha256=5c0e91…a7
Content-Type: application/json

{
  "event": "check.down",
  "occurredAt": "2026-09-22T03:14:08Z",
  "impact": "Blocks Customer export (09:00 UTC): must finish by 08:54 UTC.",
  "check": {
    "name": "Nightly database backup",
    "status": "down",
    "tags": ["prod", "db"]
  }
}

One line in your crontab

Next stop Pricing

Add a curl after the command you already run. Report an exit code to mark failures straight away, or manage checks and launches from scripts and CI with the REST API.

# Back up at 03:00, then check in
0 3 * * *  /opt/backup.sh && curl -fsS hb.example/ping/k7r2x9

# Always ping; a non-zero exit code marks the check down
0 3 * * *  /opt/backup.sh; curl -fsS hb.example/ping/k7r2x9/$?

# /start begins the timer; the next ping ends it
curl -fsS hb.example/ping/k7r2x9/start
/opt/backup.sh && curl -fsS hb.example/ping/k7r2x9

# List your checks with their status
curl -H "Authorization: Bearer hb_…" hb.example/api/v1/checks

Pricing

Both plans include every feature on this page. The only difference is how many jobs you can watch.

Free

20 checks

Create account

Pro

Coming soon

1,000 checks

Not on sale yet. Start on Free today; nothing you set up is lost when Pro arrives.

Deadlines, launches and rehearsals
Live monitor wall
Public status page with 90-day uptime
Slow-run detection
Slack, Discord and signed webhook alerts
REST API