Report
Write a compact markdown analytics report that answers the user's question in $ARGUMENTS. This skill is the high-level wrapper: pick the right sub-query (recipe or raw TQL), run it, and turn the numbers into a narrative.
Process
- Parse the question in
$ARGUMENTS. Identify: metric (count vs. users), dimension (if any), window, comparison target. - Pick the command:
- DAU/MAU trend →
tdq dau/tdq mau - Breakdown by one dimension →
tdq groupby <dim> - Which events fire →
tdq events - Funnel / retention / multi-dim groupby / derived metric → write raw TQL and run via
tdq query -(consultskills/analytics/tql/index.mdfor syntax).
- DAU/MAU trend →
- Run it — use the Bash tool. Capture the table output.
- Write the report using this structure:
## <question restated as a header>
**Top-line:** <one sentence with the key number and window>
**Method:** `<the exact `tdq …` invocation you ran>`
**Result:**
<markdown table — paste the CLI output directly>
**Interpretation:** <1–3 sentences: what the numbers mean, any trend or outlier, compared-to-what>
**Caveats:** Opt-in cohort only (TelemetryDeck samples self-selected users). <add channel skew / partial-period / small-sample notes if applicable>
- Where to save: inline in the conversation by default. If the user asks for a file, ask them where to save it; write to the absolute path they give. Suggest
~/Documents/TelemetryDeck/YYYY-MM-DD-<slug>.md. Never create new top-level folders in the current working directory.
Defaults
- Window: last 30 days unless the question implies otherwise.
- Format:
table(markdown-ready). - Always include the opt-in-cohort caveat.
- For "trending" / "growing" / "vs. last period" phrasings, add
--compare prior-period.