Niches · Product analytics & experimentation
Set up, send and verify event tracking
Two or three companies' docs show a full fit.
What counts as done
The expected events appear in the project with the right properties and identity, confirmed by reading the newest events through the API or MCP.
Measured, for our own tests: they appear within 5 minutes of sending. No docs we read state these figures, so they don't count against a listing here.
How agents ask for it: is this event being tracked
, why isn't signup showing up
, add tracking for the new feature
, audit our identify calls
, send this backend event
Does the job 3
The docs show it meets the test above, and nothing we read says otherwise.
| Listing | Why |
|---|---|
| Matomo B InnoCraft Limited | The Tracking HTTP API sends events from any backend, and Live.getLastVisitsDetails returns the latest raw visits with their events, filterable by segment, but the docs state no ingestion latency against the 5-minute bar. source |
| RudderStack MCP C RudderStack, Inc. | get_live_events polls real-time source, transformation and destination events, get_source_event_properties shows observed property types, and Tracking Plan violation tools flag schema mismatches. source |
| Statsig BB Amplitude, Inc. | Log Custom Events sends server events and GET /console/v1/events returns recently logged events in a rolling window, with /events/{eventName} for one event, but ingestion latency isn't stated against the 5-minute bar. source |
Does the job, with a limit 11
It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.
| Listing | Why |
|---|---|
| Amplitude BB Amplitude, Inc. | HTTP V2 sends events and /api/2/useractivity returns a user's events from the most recent, but it is per user, not filtered by event name, and no ingestion-to-query latency is stated against the 5-minute bar. source |
| Countly E Countly Ltd | /o/export/db reads drill_events with a Mongo filter and sort {ts:-1}, and sdk_logs_list shows raw SDK payloads by device_id, but the MCP log tool has no sort and no ingestion latency is stated. source |
| Fullstory B Fullstory, Inc. | v2 Create Event sends a server-side event, List Sessions returns a user's most recent sessions by uid or email, and get_session_events gives a chronological transcript, but no ingestion latency is stated and there is no newest-events feed by event name. source |
| LaunchDarkly BB Catamorphic, Co. (dba LaunchDarkly) | search-timeline-events returns one session's track and identify calls in time order and get-metric-events lists metric events, but verification is per session through observability, not a newest-events feed by event name, and latency isn't stated. source |
| Mixpanel B Mixpanel, Inc. | Raw Event Export filters by event and where over an inclusive date range, and Get-Issues reports data quality, but export has no sort parameter, is limited to 60 queries an hour, and no ingestion latency is stated. source |
| Mostly Good Metrics E Mostly Good Metrics | POST /v1/events ingests from any backend with plan rate limits (1,000 a minute on Free), but no documented tool returns the newest raw events; list_events and list_event_types cover names. source |
| OpenPanel B OpenPanel AB | The Track API ingests events and query_events and /export/events return raw records with properties, but no sort parameter is in the spec; the arbitrary-order bug (acc-36) was closed as completed on 2026-09-04 and not verified here. source |
| Pendo E Pendo.io, Inc. | visitorActivity returns one visitor's chronological timeline over up to 31 days and searchEntities orders Track Events by recency, but the MCP sends no events and the Track API wasn't read. source |
| PostHog B PostHog Inc. (formerly Hiberly Inc.) | execute-sql can read the newest events with ORDER BY timestamp, and a user debugged an untracked event over MCP (acc-02), but harness subagents lacked the MCP tools (acc-16) and ingestion latency isn't stated. source |
| Userpilot B Userpilot, Inc. | The Track Event HTTP API records events in real time, and the MCP can read a user's recent activity with profile permission, but no tool returns the newest events by name and the read path is undocumented. source |
| Woopra D Woopra, Inc. | Server-side track and identify calls are documented, and /profile/actions returns one profile's action timeline by email or id, but there is no newest-events feed by event name and latency isn't stated. source |
Doesn't do this job 5
Listings in the category that do something else, with the reason.
| Listing | Why |
|---|---|
| Flagsmith B Bullet Train Ltd | Exposure and conversion events are recorded for experiments against a connected warehouse; no endpoint or tool reads back the newest tracked events. source |
| GrowthBook BB GrowthBook, Inc. | GrowthBook reads events from the team's warehouse; it has no ingestion endpoint or newest-events read of its own beyond the read-only SQL tool. source |
| Harness Feature Management & Experimentation BB Harness Inc. | fme_event_type only lists event types seen in the last 30 days; live tail is a UI feature, and no endpoint returns the newest events. source |
| Heap D Content Square, Inc. | Server-side Track and Bulk Track send events to identified users, but verification is through the Live data feed in the UI; no API reads back the newest events. source |
| Optimizely Experimentation B Optimizely North America Inc. | Events export arrives daily, the next day by 13:00 UTC; no endpoint reads back the newest events. source |
What stops agents finishing
- no endpoint returns the newest events in order
- subagents in a harness lack the MCP tools
- client capture slows the app and the fix is not discoverable
- event volume limits on the free cloud tier
- no SDK for the backend's language
Around it
Before: code.edit-repo. After: product-analytics.metric-question.
Where this comes from
Weight 9 of 100 in Product analytics & experimentation: 5 of 55 account links; vendors (PostHog wizard, Amplitude sdk-setup skill, RudderStack MCP) put setup and debugging forward
The same as data: /api/v1/use-cases/product-analytics.instrument-verify.json, or the MCP tool get_use_case with id product-analytics.instrument-verify. Other jobs in this category: /api/v1/niches?category=product-analytics.
For companies
Do agents find, use and choose your tools?
An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.