Skip to content

Product demo

A product demo from signup to pull request.

This product demo follows a GTM team’s growth question through Skene’s own workspace to the fix you review. Every screen shows real product output, including the five events Skene found missing in its own code.

Start freeRead the quickstart

Follow one question through the product.

Captured from the live app on 29 October 2026. Redrawn screens keep the app’s text.

One question, six steps Select a step to inspect

Start with your email. One source is enough.

A magic link signs you in, with no password or card. Connect GitHub or Supabase and the first analysis is free. A Supabase auth schema alone gives 11 events.

Events

0 code events11 database events

Database
EventSourceFound atStatus
email_change_requestedSupabaseauth.usersCan be tracked (DB)
email_changedSupabaseauth.usersCan be tracked (DB)
email_confirmedSupabaseauth.usersCan be tracked (DB)
The first 3 of 11 rows on the Events page, Demo workspace. Redrawn.

Ask the question in plain words.

The Agent page takes the question your GTM team brings you, not a query. Six teams each get a set of questions to start from.

  • Can we tell whether acquisition events are tied to a known user or anonymous visitor?
  • Can connected analytics sources support attribution by provider and source role?
  • Do ingested analytics events carry the properties needed to identify an acquisition path?
Marketing ops questions on the Agent page in skene-web. Redrawn.

Skene names what’s missing.

Asked how to raise retention, Skene took 14 investigation steps and named the five events the answer needs, with reasons. All five were missing from Skene’s own code.

AgentI want to increase retention with our customers

14 investigation steps

To answer this question we need the following events:

EventWhyParametersStatus
first_valueThis marks when an account reaches the product outcome, which is the key point to measure retention from.plantimestampNot found
feature_usedThis shows which parts of the product customers keep returning to after activation.featureplantimestampNot found
onboarding_step_completedThis identifies where customers stop during setup, which helps find early drop-off that hurts retention.steptimestampNot found
subscription_cancelledThis captures churn directly so you can measure how retention changes over time.plantimestampNot found
support_ticket_openedThis helps connect support issues to later churn or repeated use patterns.timestampNot found
Open coverage
A saved Agent answer in skene-web. Redrawn.

Hand the fix to your coding agent.

How to fix turns the gap into a brief for a coding agent: where each call goes and what it sends.

How to fix

Recommended fix

Add analytics calls for the smallest set of events required to answer the retention question, and include every required property exactly as listed:

  1. first_value
    • Add in the setup or onboarding flow where an account reaches the product outcome.
    • Send plan and timestamp.
  2. feature_used
    • Add at the point of feature interaction.
    • Send feature, plan, and timestamp.
The first two of five How to fix items. Redrawn.

Ask Skene from your editor.

Cursor connects with a workspace API key and one configuration block. Your agent then gets 11 Skene tools, including a tracking review before you push.

Skene MCP
  1. Add the connection to your clientCursorManual setup

    Merge into your Cursor MCP configuration.

    Copy configuration
    {
  "mcpServers": {
    "skene": {
      "url": "https://www.skene.ai/api/mcp",
      "headers": {
        "X-API-Key": "YOUR_WORKSPACE_API_KEY"
      }
    }
  }
}

    Replace YOUR_WORKSPACE_API_KEY with your key before saving.

The connection step on the Skene MCP page. Redrawn.
Result of skene_analytics_review: PASS. It reviewed a working-tree diff of four files, three layout padding changes and a new schema context file, and found no analytics tracking regressions.
skene_analytics_review over MCP on a working-tree diff. Screenshot, cropped.

Catch the break on the pull request.

With PR reviews on, the Skene GitHub App checks every diff. Here it flags watch_demo renamed to demo_watch and suggests restoring the old name.

Skene's GitHub bot commenting on a pull request. It flags a tracking regression in src/lib/analytics.ts where a watch_demo event was renamed to demo_watch and offers a suggested change to restore the original event name.
A Skene GitHub App review on a pull request. Screenshot.

Start with the free audit.

Create a workspace and connect a source. One free audit per workspace, no card required.

Start free