← ClaudeAtlas

lamina-quantitative-validationlisted

Metrics discipline — use only real user-provided data; suggest what to measure post-launch. No invented A/B results.
aryaniyaps/lamina · ★ 114 · Code & Development · score 80
Install: claude install-skill aryaniyaps/lamina
# Metrics Discipline (agent-native) Agents **do not invent** analytics. Quantitative validation is **advisory** until real data exists. ## Allowed - Suggest metrics tied to workflows (completion rate, error rate, time-to-ticket) - Define success signals in `decisions.md` for post-launch - Incorporate user-provided dashboards or experiment results into scope ## Forbidden - Fabricated A/B outcomes - Made-up conversion rates - Claims of statistical significance without user data ## Verify tie-in Use **actor walk success/fail counts** as qualitative proxy pre-launch — not population statistics. ## Related - [Feedback Loops](../lamina-feedback-loops/SKILL.md)