Analytics and tracking skills
Analytics and Tracking Skills-md for AI Agents
Analytics and tracking skills-md are SKILL.md files that teach Claude, Codex and other AI agents to collect trustworthy behavioural data: a clean event tracking plan, campaign links that classify correctly, and the QA that catches broken tracking before it reaches a report. Each one carries decision rules, real failure modes and a verification step, so the data you analyse later can be trusted.
6 of 6 skills published, version 1.0.0.
Where to start: begin with event-tracking-plan, the contract that names and types every event. Deploy it through gtm-container-management and configure the property with ga4-measurement-setup, then protect it with consent-mode-tracking. Keep campaign data clean with utm-campaign-tracking, and confirm the tracking fires with analytics-qa-validation. Install the whole set to stand up trustworthy analytics, or one skill at a time for a single problem.
Event Tracking Plan
Designs an event tracking plan as a versioned data contract: one naming grammar and casing per destination, every property typed with allowed values and a null rule, an explicit identity contract, and each event mapped to its warehouse table.
UTM Campaign Tracking
Governs campaign URL tagging and source/medium data quality: one lowercase grammar, a closed list of mediums that map to the channel rules, no double-tagging against auto-tagging, and never a UTM on an internal link.
Analytics QA and Validation
Validates instrumentation end to end: pre-release QA in DebugView, schema and type checks at ingestion, production monitoring on the five observability pillars, and a decision tree that separates a real tracking fault from a consent or modelling effect.
GA4 Measurement Setup
Configures a GA4 property from an agreed plan in a fixed order: the least custom event category for each interaction, every reported parameter registered before release, 14-month retention, clean referral and identity settings, and personal data kept out at the source.
GTM Container Management
Builds and governs a Google Tag Manager container: one naming grammar, data-layer triggers instead of brittle DOM selectors, one tag per job with no duplicates, consent initialisation ordered first, and a human gate on every publish.
Consent Mode Tracking
Configures consent-aware measurement so tags respect the user choice: the Consent Mode v2 signals, the default set before any tag, basic versus advanced mode, persistence and withdrawal, server-side consent, and the legal call left to a human.
No skill matches that search. Clear it or choose All.
What analytics and tracking skills do for an AI agent
A general model can add tracking code, but left alone it guesses: it invents an event name that already exists under another spelling, tags an internal link and overwrites the real traffic source, or ships instrumentation nobody checks until a report looks wrong. These SKILL.md files give the agent the decision rules for each job, so it designs a plan, tags campaigns and validates the data the way a careful analytics engineer would, and tells you what it changed.
Where to start
Begin with event-tracking-plan, because naming and typing come before anything else. Add utm-campaign-tracking to keep source and medium data clean, since a single mistagged link can reattribute weeks of traffic. Then use analytics-qa-validation to check the tracking fires correctly before release and to monitor it in production. Install the set to cover design, collection and validation, or one skill at a time for a single problem.
What the agent needs from you
Point the agent at the destinations your data flows to, the decisions it must support, and any existing events or campaign links. Each skill asks for the specifics it needs, such as the warehouse the events land in or the approved list of campaign mediums, rather than guessing and leaving you with data that cannot be trusted.
Frequently asked questions
What are analytics and tracking skills for AI agents?
They are SKILL.md files that make an agent competent at the data side of analytics: designing an event tracking plan, governing campaign URLs, and validating that tracking works. They focus on data quality and trust, not on marketing strategy, so the numbers you analyse are correct.
Do I need GA4 or a specific tool?
No. The skills are tool-agnostic at their core and cover GA4, Segment, Amplitude, Mixpanel, Snowplow, Avo and Google Tag Manager in fenced appendices. You tell the agent your stack and it applies the rules to it.
Which tracking skill should I start with?
Start with event-tracking-plan. A clean, typed, versioned plan is the foundation, and both campaign tagging and QA depend on it. Add the other two once the plan exists.
Will these replace my analytics team?
No. They make an agent do the careful, repetitive parts to a known standard and flag anything destructive for a human to confirm. Judgement about what the business needs to measure stays with you.
Sources and more data skills
Each skill lists its own sources with the date it was last checked, drawn from the official documentation of GA4, Google Tag Manager, Segment, Amplitude, Mixpanel, Avo and Snowplow, and from standard practice. For the rest of the data library, see the data-analysis, sql-querying, data-cleaning-preparation and dashboards-reporting subcategories.
