PostHog Growth Workspace

Contribute within authority. The active workspace contract and current task scope decide whether a run may modify this skill, open an upstream pull request, or record portable growth-tooling friction. When authorized, fix the canonical source and use its normal release flow. When upstream maintenance is not authorized, leave the skill untouched; if the normal .growth/log.md handoff is already in scope, record concrete, non-duplicate friction there. Invoking this skill never implies product-development scope.

Run a durable growth operating workspace on top of a product's live PostHog data. The goal is not advice; it is evidence from real events, one verified action per pass, and continuity between sessions.

Use this skill as a mode router. Load only the reference the selected mode needs.

Ground Rules

• Every claim about the product cites live PostHog data (a query, insight, or replay), never intuition. If the data disproves the hypothesis, that is the finding.
• PostHog access is CLI/REST only: scripts/pg-query.mjs (HogQL + raw API) or posthog-cli. Auth comes from the environment (POSTHOG_PERSONAL_API_KEY); never print keys.
• Keep one current focus ticket at a time; task lists live only in .growth/backlog.md.
• Evidence goes to .growth/audit.md, decisions to .growth/strategy.md, handoffs to .growth/log.md, dated reviews to .growth/reports/.
• Registries are rewritten in place, never forked: dashboards in dashboards.md, experiments in experiments.md, campaigns in campaigns.md.

Boundaries

Three-way contract — respect it in both directions: