← ClaudeAtlas

metrics-anomaly-investigatorlisted

Investigate metric anomalies — build a ranked hypothesis framework and stakeholder narrative from a plain-language description. Use when a dashboard number drops or spikes unexpectedly. Triggers: 'investigate metric anomaly', 'my metric dropped', 'расследуй аномалию метрики', 'у меня упала метрика'.
KirKruglov/claude-skills-kit · ★ 9 · AI & Automation · score 82
Install: claude install-skill KirKruglov/claude-skills-kit
# Metrics Anomaly Investigator This skill takes a plain-language description of a metric anomaly (metric name, time period, delta, and any available context) and produces a structured investigation framework with ranked hypotheses and a ready-to-send stakeholder narrative. No database access or code required — designed for product managers working in Cowork. **Input:** - Free-form description of the anomaly: metric name, observed period, magnitude of change, and any context (recent releases, campaigns, external events) **Output:** - Markdown response with Anomaly Summary, Investigation Framework table, Priority Checks, and Draft Stakeholder Narrative --- ## Language Detection Detect the user's language from their message: - If Russian (or contains Cyrillic): respond in Russian - If English (or other Latin-script language): respond in English - If ambiguous: respond in the language of the trigger phrase used --- ## Instructions ### Step 1: Validate Input 1. Check that the user provided at minimum: - A metric name (or description of what was measured) - An observed change (delta, direction, or time period) - If either is missing: stop and respond: "To investigate the anomaly I need at minimum: (1) metric name, (2) time period, (3) observed change. Please provide these details." Do not generate a framework. 2. If the input is vague (e.g., "our metrics are bad" with no specific anomaly): - Ask the user to identify one specific metric and its observed change