Alerts

From the map

Select a price node or asset and choose Set alert from its info card. For a place rather than an asset, drop a pin and create a radius alert. Multi-select or lasso selection can carry a chosen set of nodes into one rule.

1. 1Open a live grid and select an asset, node, pin, or group.
2. 2Choose Set alert from the contextual action.
3. 3Review the scope at the top of the right-side alert panel.

From Radar

Use a widget's alert action to monitor the metric in place. The grid, metric, current value, and a sensible starting threshold are added to the draft automatically.

From an asset page

Use Set alert on an individual asset page when you want the rule permanently tied to that facility or settlement point.

From the Alerts manager

Open Alerts in the main navigation to start from a live template: nodal price, local congestion, day-ahead versus real-time spread, grid events, low reserve headroom, or forecast divergence. Asset and area rules carry their targets in from the map.

From an MCP client

A connected AI client can list, inspect, create, and archive your alerts through Ask the Grid MCP. Creation requires an explicit request and the separate alert-write OAuth permission. The client should compare unusual thresholds with recent data and ask for confirmation instead of creating a rule that is almost always or almost never true.

1. 1Connect the global MCP server or the endpoint for one ISO.
2. 2Reconnect and approve alert-write access if the connection predates MCP alert management.
3. 3Explicitly ask the client to create or archive the alert.

Fixed threshold

Trigger when a supported metric goes above or below a number—for example, nodal LMP above $100/MWh or ERCOT system load above 75,000 MW.

1. 1Choose the metric and Above or Below.
2. 2Enter the threshold in the metric's displayed unit.
3. 3Choose how much evidence is required before the rule activates.

Percent change

For signals with executable history, compare the latest value with the start of a 15-minute to 24-hour lookback. The editor only offers this condition for market, metric, and target combinations backed by historical evaluation.

Day-ahead versus real-time spread

DA/RT spread alerts compare the two prices as an absolute difference in $/MWh. A rule such as real time 20% above day-ahead is not supported; provide a dollar spread such as DA/RT above $50/MWh instead.

Moving average

For signals with executable history, trigger when the latest value crosses its trailing average over the chosen lookback. No separate numeric threshold is required.

SPP, MISO, and ISO-NE coverage

SPP supports exact settlement-location real-time LMP, day-ahead versus real-time spread, system load, and load-versus-forecast divergence. MISO supports exact pricing-node LMP and spread, system load, operating headroom calculated as committed capacity minus realized demand, load-versus-forecast divergence, published resource-adequacy surplus, and reported planned plus forced generation outages. ISO-NE supports exact location LMP and spread, system load, load-versus-forecast divergence, and the next-day report's resource-adequacy surplus and planned plus forced outage or reduction totals. Price, load, and MISO operating-headroom series support fixed, percent-change, and moving-average conditions. The publication-based forecast, adequacy, and outage metrics support fixed thresholds only.

Lookback and confirmation

Lookback controls the evidence window. Confirmation controls how many consecutive evaluations must match before the alert activates. Increase confirmation for volatile metrics to avoid reacting to a single interval.

Recovery

Recovery defines when an active alert returns to normal. Use a separate recovery threshold when the metric tends to hover around the trigger; this prevents repeated activate/recover notifications at the boundary.

Grid events

Watch for outages, operator notices, reliability actions, and other supported operational events. Source-backed event adapters currently cover ERCOT and NYISO; SPP, MISO, and ISO-NE event rules do not appear in the editor until their operator notices can be mapped without inventing event identity or status. Scope a supported rule to a grid, asset, selected set, or geographic area, then filter the event categories and severity that matter.

Asset-scoped events

When an event source identifies a supported node, resource, line, zone, or DC tie, an asset rule can watch that exact subject. The notification identifies the subject and new event evidence rather than reporting only a grid-wide occurrence.

Pin and radius

Drop a pin, choose a radius, and watch for supported events near that location. The radius stays visible while you configure the condition so the monitored footprint is clear.

Multi-select

Select several nodes to monitor them as a fixed set. This is useful for a portfolio, corridor, development screen, or operating footprint that does not form one simple radius.

Lasso selection

Use the map lasso to save a frozen polygon for supported event alerts. Events with spatial evidence inside that boundary may match later; value alerts still require concrete supported node or grid series and are not aggregated across arbitrary geometry.

Manage the beta

Alerts is enabled by default for eligible accounts. After signing in, use Settings → Experiments to turn it off or back on; the preference follows your account across devices.

1. 1Create or sign in to a free account.
2. 2Open Settings, then choose Experiments.
3. 3Confirm Alerts is on, or change the preference.

Choose where alerts arrive

A free account is required because an alert runs after you leave the page and needs a durable destination. Account email is ready automatically. Save SMS, Slack, or Microsoft Teams once under Alerts → Destinations, test the connection there, and then select any combination of saved destinations on each rule.

1. 1Confirm Alerts is enabled for your account.
2. 2Open Alerts → Destinations and choose Add destination.
3. 3Name the destination and enter an E.164 phone number or incoming webhook URL.
4. 4Verify an SMS number with its one-time code, or send a test to a saved Slack or Teams webhook.
5. 5Select that ready destination under Notify when creating or editing a rule.

Control notification volume

Value rules are episode-based: a true condition does not notify on every one-minute evaluation. Each rule also has a rolling 24-hour cap for each selected destination, covering repeated episodes or bursts of distinct grid events. Ask the Grid keeps recording triggers in Activity after a cap is reached, so delivery protection never removes evidence.

1. 1Choose Per destination / 24h under Notify.
2. 2Use a lower limit for noisy exploratory rules and a higher limit for critical event streams.
3. 3Review capped triggers in Alerts → Activity.

Webhook credentials stay private

Slack incoming webhooks and Microsoft Teams workflow webhooks are validated and encrypted when you save the destination. Read APIs return its name, safe target hint, health, and usage count—not the number or webhook. Rotating the target updates future occurrences for every linked rule; deliveries already queued retain their original encrypted snapshot for safe retry.

1. 1Create an incoming webhook in the Slack or Teams channel you control.
2. 2Paste it into a named destination and save it.
3. 3Use Test in Alerts → Destinations to confirm the connection.

Reuse and remove destinations safely

One saved destination can serve many alerts, including several destinations of the same channel. Alerts shows how many active rules use each one. A destination in use cannot be removed; edit those rules first so disconnecting a service never silently changes delivery. The built-in account-email destination follows your current account address and cannot be removed.

Automate through the developer API

The Alerts API exposes destinations as first-class resources at /api/v1/alert-destinations. Create or list destinations, use their IDs in an alert's destinationIds field, rotate credentials, test one service, and inspect safe health metadata with the same API key used for alert rules. SMS uses explicit /verification/request and /verification/check operations; the API stores consent and verification timestamps, never the submitted code.

Why sign-in is required

Anonymous visitors can inspect the public grid, but persistent alert state, delivery destinations, and evidence history belong to an account. The signup gate shows the feature in context and returns you to the alert draft after authentication.

Use the three workspace tabs

Activity is the notification feed: filter it by all, unviewed, or viewed activity and clear the current unread queue at once. Triggers is the rule manager: filter by state, pause a rule, or open it to review evidence and delivery history. Destinations is the account-level service area for email, SMS, Slack, and Microsoft Teams connections, defaults, health, tests, and usage. It remains available before the first trigger is created, while New alert opens the builder without adding another tab.

Understand rule health

Healthy means the evaluator can read the source and run the condition. Degraded or error states describe source or evaluation trouble separately from whether the monitored condition is active.

Read evidence history

Recent activity combines transitions across every rule into one chronological feed. Expand a rule for its focused evidence history. Both views retain observed time, source, value when available, and delivery status—including triggers recorded after an email cap was reached.

See alert activity in Radar

Radar → Events includes an Ask the Grid alerts widget filtered to the current ISO. It sits beside the operator event stream, so a user can compare what the grid reported with which personal rules actually triggered.

Pause, clone, or archive

Pause stops evaluation without deleting the rule. Clone creates a paused copy for safe editing. Archive stops evaluation and removes the rule from the active manager while retaining service-side evidence.

Missing data

Choose how the rule should treat a delayed or missing source. Missing data is not automatically normal and should not silently resolve an active condition.