Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.sourcemedium.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

Build BI dashboards from SourceMedium warehouse data with correctness first — the agent discovers your data, locks down metric definitions, validates SQL, then renders the dashboard. Default output is a portable standalone HTML file that opens in any browser.

​
Prerequisites

The install command requires npx, which comes with Node.js 18+. Run node -v to check. If you see “command not found,” install Node.js from nodejs.org first. Install SM BigQuery Analyst first if you need setup verification or access troubleshooting — the dashboard builder assumes BigQuery access is already working.

​
Install

npx skills add source-medium/skills --skill sm-dashboard-builder
Install both skills together: 
npx skills add source-medium/skills --skill sm-bigquery-analyst
npx skills add source-medium/skills --skill sm-dashboard-builder
If your agent does not support automatic install through the skills CLI, copy the skill folder into that agent’s configured skills directory.

​
Update

If you installed an earlier copy, update both skills before building dashboards: 
npx skills update sm-bigquery-analyst -y
npx skills update sm-dashboard-builder -y

​
Quick Start (Copy/Paste)

Copy the block below and paste it into your coding agent to build your first dashboard.
Use the SourceMedium Dashboard Builder skill to create an executive overview
dashboard for my project sm-[your-tenant-id].

Include:
- Net revenue, orders, and AOV scorecards for the last 30 days
- Revenue trend by day
- Revenue breakdown by channel

Discover available tables and validate all queries before building.
Produce dashboard.html with SQL receipts and a dashboard_manifest.json.

​
What It Does

Metric definitions before SQL — before writing any query, the agent defines each metric: the formula, the source table, the time level (daily, per-order, per-customer), and for ratios like AOV or ROAS, the numerator and denominator. This prevents the most common BI mistake — summing a rate metric across rows and getting a wrong answer. BI-safe SQL — SELECT-only, fully qualified tables, bounded time filters, dry-run validated, one query per dashboard tile. Most queries cost well under a dollar at standard BigQuery rates. Manifest-driven output — builds a dashboard_manifest.json that records every tile’s metric definition, SQL receipt, row count, dry-run bytes, and QA status. The manifest is validated before rendering, so errors surface before the dashboard is built. Portable HTML dashboard — dashboard.html embeds chart specs, query results, SQL receipts, and QA notes. Open it in any browser — no server, no login required. BI tool handoff — produces Metabase SQL cards, Looker Studio queries, or app-ready data contracts when a BI tool is the target instead of standalone HTML. No guessed metrics — if a requested metric isn’t available in your data, it appears as a note in the dashboard rather than a substituted or fabricated number.

​
Example Prompts

​
Executive overview dashboard

Build an HTML executive dashboard for the last 30 days.

Tiles:
- Scorecards: net revenue, orders, AOV, new customers, ad spend, MER
- Trend: net revenue and orders by day
- Breakdown: revenue by channel

Discover available tables and stores first. Validate all queries.
Output dashboard.html and dashboard_manifest.json.

​
Marketing performance dashboard

Build a marketing performance dashboard for the last 30 days.

Tiles:
- Scorecards: ad spend, platform ROAS, blended MER, CAC
- Trend: spend and platform-reported revenue by platform
- Breakdown: campaign type performance for each platform

Note: TikTok GMV Max campaigns should be separated from standard campaigns.

​
Metabase handoff

Create Metabase-ready SQL cards for CAC, MER, ad spend, and revenue by
channel for the last 30 days. Include field definitions, expected
visualization type, and layout order.

​
Hybrid dashboard (your data + SourceMedium)

Build a dashboard that joins my inventory table
[your-project].[your-dataset].inventory_snapshots to SourceMedium orders.

Show:
- Revenue by SKU
- Sell-through rate by product category
- Inventory days-on-hand for top products

Document the join keys and run a cardinality check before joining.

​
What You Get

OutputDescription
dashboard.htmlOpen in any browser — no server required. Embeds chart specs, query results, and SQL receipts.
dashboard_manifest.jsonSource of truth for the dashboard: metric definitions, SQL receipts, row counts, and QA status per tile.
SQL receiptsOne copy/paste BigQuery query per tile, with dry-run byte estimate and freshness check.
Notes sectionMetric definitions, data caveats, freshness date, and QA status — visible in the dashboard footer.

​
After Building

​
View your dashboard

Open dashboard.html in any browser — Chrome, Safari, Firefox, Edge. If your agent built it in a project folder, look there. You can share the file directly or host it anywhere that serves static files.

​
Validate before sharing

Run strict validation before sharing any dashboard externally: 
Validate dashboard_manifest.json in strict mode before I share this.
The agent will check metric definitions, SQL safety, row counts, dry-run bytes, ratio denominator checks, and chart field matching.

​
Update a dashboard

Update dashboard.html to add a product revenue breakdown tile for the
top 10 SKUs by net revenue last 30 days.

​
Port to Metabase

Convert dashboard_manifest.json to Metabase SQL cards and a layout plan
I can recreate in Metabase.

​
Debug a tile

If a chart looks wrong: 
The revenue trend chart looks off. Show me the SQL receipt for that tile,
rerun the query, and compare the result to the revenue sanity check query.
If the numbers still look wrong after debugging, share the SQL receipt and a screenshot of the chart with support.

SM BigQuery Analyst

For setup verification, raw analysis, and hybrid data joins.

SQL Query Library

SourceMedium SQL templates and patterns.

Metric Catalog

180+ pre-defined metrics with calculations.

Table Docs

Schema-level documentation for core tables.