Stallguard

Docs

How Stallguard works

Three ideas: a sourceis where a client's results land, a monitor watches that they keep arriving, and an alert reaches you the moment they stop.

Step 1

Connect a source

A source is read-only. Two kinds today:

  • Airtable: create a read-only personal access token (scope data.records:read) and paste it in.
  • Google Sheets: share the sheet “anyone with the link can view” and paste the spreadsheet ID. No token needed.

Step 2

Build a monitor

A monitor watches one of two things:

  • A data table(Method A): point it at a source's table or sheet. We watch the row count grow, or the freshnessof a timestamp column. Zero setup on the client's side.
  • A heartbeat(Method B): for automations that don't land in a table (send email, post to Slack, anything). Add one HTTP step at the end that calls the ping URL we give you. Works with Zapier, Make, n8n, or anything that can make a request.

For both you set how often results should arrive and a grace period. Example: every 15 minutes, grace 5 → we alert if nothing new for 20 minutes.

Statuses

What the colors mean

  • Healthy: data is arriving on schedule
  • Warning: later than expected, inside the grace window
  • Silent: past expected + grace with nothing new, so an incident opens
  • Unknown: we couldn't check (e.g. expired token), never confused with silent

Step 3

Get alerted

Add an email address or a Slack webhook in Settings. When a monitor goes silent we notify every channel, and again when it recovers. One alert per incident, never a stream.

FAQ

Common questions

Do you write to my client's data?

Never. Stallguard is strictly read-only. See Security.

Do you store my client's records?

No, only the measurements we need to detect silence (a count and the latest timestamp), not the record contents.

Can you watch private Google Sheets?

Today we read sheets shared by link. Private-sheet access via Google OAuth is on the roadmap.

How often do checks run?

Automatically, on a schedule. You don't click anything. The cadence you set per monitor decides when it's considered late.