---
description: End-to-end read-only tracking audit with a scored report and a prioritised fix list
argument-hint: [site URL or journeys or HAR files]
---

Use the tracking-audit skill for this task: $ARGUMENTS

Read martech-context/ first if it exists, especially stack.md, tracking-plan/, decisions.md and naming-rules.md.
If $ARGUMENTS is empty or vague, ask for the site, the journeys that matter (lead form, signup, checkout), the regions served and the destinations in use.
If $ARGUMENTS names HAR files, ask which scenario (no-choice, reject, accept or journey), region and run each file is before running anything.
If the consent model for a region in scope is not documented, stop and ask for it from the privacy owner.
Never ask for passwords, tokens, API secrets or service account keys in chat.
Fill the scope sheet before capturing anything. Banner clicks, synthetic click IDs on live pages, test leads, test signups, test orders and platform test events each need a named yes.
Drive scenarios in an agent-owned isolated browser, and run the scripts on HAR files exported from DevTools or Playwright: Chrome DevTools MCP cannot save a HAR, and Playwright MCP records one only when recordHar is set in its config. Never use the user's everyday browser profile.
Follow the skill's procedure in order and finish with its acceptance checks.
Stay read-only. Never publish, change settings, upload conversions, or send Measurement Protocol, Conversions API or Events API calls.
Hand back the scored report, the evidence ledger, the not-checked list and a prioritised fix list a colleague can action.

If the tracking-audit skill is not installed, install it with: npx skills add https://www.martechdevs.com --skill tracking-audit -a claude-code