Conditions

A condition tells Alertee what the healthy result of your SQL query looks like. The check passes while the condition holds and alerts when it is violated — the same polarity for single-value and grid checks.

Standard conditions

These apply to queries that return a single value.

Condition	Healthy when…	Value field
`greater_than`	result > threshold	`threshold`
`less_than`	result < threshold	`threshold`
`greater_or_equal`	result ≥ threshold	`threshold`
`less_or_equal`	result ≤ threshold	`threshold`
`equals`	result = value	`threshold` (or `match_value` for text)
`not_equals`	result ≠ value	`threshold` (or `match_value` for text)
`between`	min ≤ result ≤ max	`min_value` + `max_value`
`not_between`	result < min or result > max	`min_value` + `max_value`
`contains`	text result contains the substring	`match_value`
`regex`	text result matches the pattern	`match_value`
`timestamp_age`	a timestamp result is no older than N minutes (freshness)	`threshold` (minutes)
`is_null`	the result is NULL	—
`is_not_null`	the result is not NULL	—
`exists`	the query returns any row	—

The condition describes the healthy state. For example, greater_than 10 passes while the value is above 10 and alerts when it drops to 10 or below. To alert when a value goes above a limit, assert the healthy ceiling instead, e.g. less_or_equal 10.

Examples

-- Healthy while failed payments stay at or below 10 in the last hour;
-- alerts when they exceed it.
SELECT COUNT(*) FROM payments WHERE status = 'failed'
  AND created_at > NOW() - INTERVAL '1 hour'
-- Condition: less_or_equal 10

-- Alert if no new users signed up in the last 30 minutes
SELECT COUNT(*) FROM users WHERE created_at > NOW() - INTERVAL '30 minutes'
-- Condition: greater_than 0   (healthy while at least one signed up)

-- Data freshness: alert if the newest row is more than 15 minutes old
SELECT MAX(created_at) FROM events
-- Condition: timestamp_age 15

-- Alert if the latest deploy status isn't "success"
SELECT status FROM deploys ORDER BY created_at DESC LIMIT 1
-- Condition: equals  (match_value: success)

Escalation thresholds

A single-value check can carry multiple thresholds, each with its own alert level — warn as a metric approaches a limit, go critical when it crosses it, without creating two checks.

The primary condition alerts at the check’s Alert Level.
Each escalation threshold is another healthy comparison with an explicit level (info, warning, error, critical).
Every threshold describes the healthy state. On each run, all of them are evaluated and the worst violated level wins.
When a worse threshold is crossed while an incident is already open, the incident escalates: its severity is raised and a fresh alert goes out at the new level (you’re not left assuming it’s still just a warning).

Example — monthly usage with a 100k hard limit:

Primary condition:      less_than 75000     (check Alert Level: warning)
Escalation threshold:   less_than 100000 →  critical

At 80k the check fails at warning. If usage reaches 100k the open incident escalates to critical and re-alerts. In the app, add levels under Escalation thresholds in the condition step; via the CLI use --extra-threshold "less_than:100000:critical"; via MCP pass thresholds. Grid checks don’t support escalation thresholds.

Custom alert messages

By default the alert/incident text is auto-generated (e.g. “Value 17 is not greater than 1560” or “Extracted value 17 from …”). Set a custom message to replace it with something your responders can act on. It works for any check type — single-value SQL, api_metric, push, grid, http, and heartbeat — and is a property of the check itself, so you can set or change it on its own without touching the condition. The message supports templating, substituted at alert time:

Token	Resolves to
`{check_name}`	The check’s name
`{value}`	The value that triggered the alert
`{threshold}`	The comparison value the check is testing against
`{name}`	The failing dimension (grid checks only)

{check_name} stale: {value} rows, expected at least {threshold}. Re-run the sink.

The custom message changes only the alert/incident body — run_check still shows the literal verdict so you can always see the raw comparison. Set it in the app’s Alerts step, via MCP create_check/update_check with custom_message (on update_check it’s settable on its own — no need to re-send the condition; pass an empty string to clear it), or via the CLI with --alert-message on checks create or checks update.

Grid conditions

Grid checks monitor multiple dimensions at once — useful when you want to watch many stores, regions, services, or endpoints in a single check. Your query must return two columns: name (the dimension) and value (the metric).

SELECT store_id AS name, COUNT(*) AS value
FROM orders
WHERE created_at > NOW() - INTERVAL '15 minutes'
GROUP BY store_id

The comparison you choose describes the healthy state of each cell — exactly like a single-value check — and the check alerts on cells that don’t meet it. Grid cells support the numeric comparisons (greater_than, less_than, greater_or_equal, less_or_equal, equals, not_equals).

Grid condition	Alert fires when…
`all_cells`	Any dimension fails the comparison (every dimension must satisfy it)
`any_cell`	No dimension satisfies the comparison (at least one must)
`dimension_missing`	An expected dimension is absent from results
`dimension_added`	An unexpected dimension appears in results

Example: every store should have at least 5 orders

SELECT store_id AS name, COUNT(*) AS value
FROM orders WHERE created_at > NOW() - INTERVAL '15 minutes'
GROUP BY store_id

Condition: all_cells — each cell’s value greater than or equal 5 This fires if any store has fewer than 5 orders in the last 15 minutes. (Use any_cell instead when only one dimension needs to satisfy the condition — e.g. “at least one replica must be caught up”.)

Getting started

Checks

Inbox

Alerts

Account

Status

Standard conditions

Examples

Escalation thresholds

Custom alert messages

Grid conditions

Example: every store should have at least 5 orders

​Standard conditions

​Examples

​Escalation thresholds

​Custom alert messages

​Grid conditions

​Example: every store should have at least 5 orders

Standard conditions

Examples

Escalation thresholds

Custom alert messages

Grid conditions

Example: every store should have at least 5 orders