Connect Claude to Klaviyo

List profiles in the Klaviyo account. Returns real people - email addresses, phone numbers, names and custom properties - so ask for a filter rather than paging the whole account. Filters use Klaviyo's expression syntax: equals(email,'[email protected]'), any(id,["01ABC","01DEF"]), greater-than(created,2026-01-01T00:00:00Z).

Fetch one profile by its Klaviyo ID, optionally with its list and segment memberships. Returns that person's contact details and properties.

List the account's lists - the manually managed audiences a person is added to or removed from, as opposed to segments, which are rule-driven. Returns list names and IDs, not their members. Klaviyo caps this endpoint at 10 per page.

Fetch one list by ID. Pass additional_fields ["profile_count"] to include how many people are on it - Klaviyo rate-limits that variant hard (about one call a second), so do not ask for it in a loop.

List the account's segments - audiences defined by rules, which Klaviyo re-evaluates continuously. Returns definitions and IDs, not the people in them. Klaviyo caps this endpoint at 10 per page.

List campaigns. Klaviyo rejects this endpoint without a channel filter, so `filter` always carries one and defaults to equals(messages.channel,'email'). Whatever you pass must still name a channel - combine with and(...), e.g. and(equals(messages.channel,'sms'),equals(status,'Sent')). Valid channels are email, sms and mobile_push.

Fetch one campaign by ID, with its status, schedule and audience references. Pass include to pull the campaign's messages - subject lines, preview text and template references - in the same call.

List the metrics Klaviyo tracks for this account - Placed Order, Opened Email, Viewed Product, and anything your integrations send. Call this first to find the metric ID that query_metric_aggregates needs.

Count or sum a metric over a date range, optionally bucketed by interval and broken down by a property. This is the reporting action: it answers "how many orders last month", "revenue by channel", "opens per day" without returning a row per person. Get the metric ID from list_metrics. Measurements are count, sum_value, unique. Break down with group_by, e.g. ["$attributed_channel"] or ["$flow"].

List the account's automated flows - welcome series, abandoned cart, win-back - with their status and trigger type. Returns definitions, not the people who went through them.

List raw events - one row per thing one person did, with the event properties that came with it. Filter by metric_id or profile_id and by datetime; without a filter this is the account's whole firehose. Use query_metric_aggregates instead when the question is a total rather than a list of individual actions.