analyze-posthog-usagelisted

# Analyze PostHog Usage Use PostHog as an evidence source, not as a fixed dashboard. Compose small, bounded queries around the user's current question, then explain what the results support, what they do not support, and what to investigate next. ## Workflow 1. Establish the question and decision. If the user is exploratory, default to recent activity, journeys, adoption breadth, and anomalies rather than asking for a full metric specification. 2. Inspect the product's current analytics source of truth. For Cradle, read [references/event-model.md](references/event-model.md) and refresh it from the referenced TypeScript source when the code has changed. 3. Choose a cohort and time window. Default to both a short window (7 days) and a comparison window (previous 7 days or 28-day context). State the choice. Keep internal usage visible but separate it from external headline metrics. 4. Read [references/query-patterns.md](references/query-patterns.md) and compose only the query blocks needed for the question. 5. Run named HogQL queries with `scripts/posthog_query.py`. Start with data availability and quality before interpreting behavior. 6. Cross-check important conclusions with a second cut, such as another time window, audience, build channel, app version, or minimum-activity threshold. 7. Report evidence, interpretation, uncertainty, and the smallest useful next action. Suggest product or instrumentation changes, but do not make them unless the user separately asks. ## Quer