From magic-powers
Deep-dive investigation of Amplitude charts to identify trends, anomalies, and root causes. Uses mcp__Amplitude__query_chart, mcp__Amplitude__render_chart, mcp__Amplitude__get_event_properties.
npx claudepluginhub kienbui1995/magic-powers --plugin magic-powersThis skill uses the workspace's default tool permissions.
- A stakeholder asks "why did this metric drop/spike?"
Generates design tokens/docs from CSS/Tailwind/styled-components codebases, audits visual consistency across 10 dimensions, detects AI slop in UI.
Records polished WebM UI demo videos of web apps using Playwright with cursor overlay, natural pacing, and three-phase scripting. Activates for demo, walkthrough, screen recording, or tutorial requests.
Delivers idiomatic Kotlin patterns for null safety, immutability, sealed classes, coroutines, Flows, extensions, DSL builders, and Gradle DSL. Use when writing, reviewing, refactoring, or designing Kotlin code.
Load the chart using mcp__Amplitude__query_chart (deep-dive, interactive) or mcp__Amplitude__render_chart (visual snapshot). Identify the primary patterns:
Always extract specific numbers. Avoid vague statements like "the metric went up." Say "DAU increased 23% from 41,200 to 50,700 between March 1 and March 15."
Scan for deviations from expected behavior:
Flag any anomaly with: when it occurred, how large the deviation was, and whether it recovered.
Break the metric down by key dimensions to isolate the driver. Apply segmentation systematically:
Use mcp__Amplitude__get_event_properties to discover available properties for segmentation. Look for the segment where the pattern is most pronounced — that is where the root cause likely lives.
Correlate the finding with external context:
Ask: "What changed on or just before this date?"
Formulate 2-3 specific, falsifiable hypotheses for what is causing the pattern. Rank by likelihood based on evidence. Each hypothesis should be in the form: "I believe X is happening because I see Y in the data, which would be consistent with Z."
State your conclusion clearly:
mcp__Amplitude__query_chart — deep-dive into a single chart with interactive datamcp__Amplitude__render_chart — visual snapshot of a chart for pattern recognitionmcp__Amplitude__get_event_properties — discover dimensions available for segmentationmcp__Amplitude__get_charts — find related charts for cross-referencingmcp__Amplitude__query_amplitude_data — run custom queries for additional contextOutput is written in narrative paragraphs, not bullet lists. The analysis reads like an analyst memo, not a database dump.
Structure:
Every claim must be backed by a specific number from the data. No vague statements.