SQL Editor, charts & dashboards

A first-class SQL workspace for analysts and data engineers - with the Colrows compiler in the loop. Every query is bound to the semantic graph, every result is reproducible, and every chart you build inherits the same governance.

SQL Editor

The SQL Editor lets you author queries with full warehouse autocomplete, schema browsing, and inline parameter binding. Queries are compiled, not just sent - when you reference a metric or concept by name, Colrows resolves it through Consensus before running anything.

  • Semantic-aware autocompleteType a metric or concept name and Colrows offers governed completions - definitions, grains, allowed dimensions.
  • Compile previewSee the resolved plan before you run - proven join path, applied constraints, dialect-perfect SQL.
  • Bookmarks & versionsSave queries with version history. Every revision is point-in-time reproducible.
  • Real-time collaborationShare a query with a teammate. Comments, presence, and shared parameter values like a doc, not a Slack paste.
  • Persona-aware executionSame query, different personas → different allowed subgraphs. No surprise data leakage.
  • Multi-source joinsJoin across datasources where the semantic graph proves a valid path. No data movement; the engine pushes down per dialect.

Charts & dashboards

Any query result can become a chart with one click. Colrows infers chart type from the shape and grain of the result, then lets you customize. Charts are always backed by their compiled plan - refreshing a chart re-runs the same governed pipeline, not a stale CSV.

  • Time-series, bar, stacked-bar, area, scatter, heatmap, sankey, and KPI tiles.
  • Cross-filtering - clicking a value on one chart filters the rest of the dashboard.
  • Parameters - date ranges, regions, products as first-class controls bound to the underlying plan.
  • Embed - drop any chart into an external app over an authenticated iframe or REST API.
  • Export - CSV, Excel, PNG, or PDF; all exports include a trace ID for audit lookup.
Promote a query into a Signal.

Right-click any saved query → "Create Signal" to schedule it on a cron, set thresholds, and route alerts. See Signals.

Authoring tips

  • Reference metrics and concepts by name (e.g. net_revenue) instead of writing the underlying expression - you'll inherit governance and stay reproducible.
  • Use the :as_of hint to pin a query to a historical semantic state for backfills.
  • Prefer WHERE on graph-defined dimensions over hand-rolled filters; the planner can prune more aggressively.
  • For long-running queries, the editor shows the compile-time cost estimate before you run, so you can refine before paying for the scan.