Event Instrumentation Spec

Event Instrumentation Spec

Use this skill to design or review product analytics instrumentation so that events are reliable, interpretable, and safe to collect.

When to Use

• You are adding analytics to a new feature.
• Existing events are inconsistent (names/properties differ across platforms).
• You need a tracking plan that engineers, data, and product can all sign off on.

Inputs

Ask for:

• Product surface(s): web/iOS/Android/Flutter.
• Analytics stack: Segment, GA, Amplitude, Snowplow, custom.
• Privacy constraints: PII policy, consent model, regions (GDPR/CPRA).
• The decision(s) events must support (funnels, retention, revenue, experiments).

Workflow

1. Define measurement goals

• State the questions to answer (not the events).
• Define success metrics and leading indicators.

1. Establish naming and property rules

• Consistent verb_noun naming (or a clear alternative).
• Property types and allowed values.
• Stable identifiers and join keys.

1. Design the event set

• Keep it minimal: instrument decision points, not every UI click.
• Define canonical events shared across platforms.

1. Add privacy guardrails

• Explicitly list which fields are PII and how they are handled.
• Ensure consent gating and deletion requests are supported.

1. Make it testable

• Define QA steps and expected payloads.
• Define backfill/rollout steps if changing existing event schemas.

Outputs

Tracking Plan Template

# Tracking Plan: 

## Goal
- Decision(s) supported:
- Primary metric(s):
- Secondary metric(s):

## Naming Rules
- Events: _ (e.g., view_pricing, submit_checkout)
- Properties: snake_case
- IDs: user_id (stable), session_id, device_id (if permitted)

## Events
| Event Name | When Fired | Required Properties | Optional Properties | Privacy Notes | Owner |
|---|---|---|---|---|---|
|  |  |  |  |  |  |

## QA
- How to verify locally:
- Sample payloads (expected):
- Monitoring after release (dashboards/alerts):

## Schema Changes
- Backward compatibility plan:
- Deprecation plan (if replacing old events):

Differentiator

This skill treats tracking as an API: versionable schemas, explicit privacy constraints, and testable payload expectations rather than ad-hoc event lists.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: zhyGuGu
• Source: zhyGuGu/agent-skillkit
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.