Data exploration & SQL assistant

Ask a question about your data and the analyst profiles it, writes SQL, verifies the results and the statistics, and drafts a sound analysis summary for you to approve.

What it installs

Agents 2

  • Data Analyst

    Writes dialect-correct SQL, validates results, and drafts an analysis summary.

  • Analysis Verifier

    Sanity-checks query grain, magnitudes, and statistical reasoning before approval.

Workflows 1

  • Explore data

    Profile, write SQL, independently verify results and statistics, draft, and hand off.

Documents 1

  • Schema reference and metric definitions

    Your editable data dictionary — tables, grain, keys, caveats, and agreed metric definitions.

Goals 1

  • Questions answered with validated evidence

    Business questions are answered with reconciled SQL and statistically honest analysis.

Skills 3

  • sql-queries

    Write correct, performant, readable SQL across warehouse dialects — CTEs, window functions, cohort/funnel/dedup patterns, and dialect gotchas. Adapted from anthropics/knowledge-work-plugins/sql-queries.

  • explore-data

    Profile a dataset before analyzing it — structure, column classification, per-column stats, quality flags, and relationships. Adapted from anthropics/knowledge-work-plugins/explore-data.

  • statistical-analysis

    Apply sound statistics — distributions, trends, outliers, significance — and flag traps like correlation-vs-causation and Simpson's paradox. Adapted from anthropics/knowledge-work-plugins/statistical-analysis.

Folders 1

  • Data Exploration

Requirements

What this template expects to do its job. Task Machine does not verify these — you decide whether your setup is ready.

  • Connected product analytics — Queries run against your connected product analytics through PostHog. The analyst reads the schema, runs read-only queries, and validates results. Until you connect it, it works from attached exports and the schema document.

Get started

Install Data exploration & SQL assistant and run it with approvals.

Join the waitlist and we will send early access when the first private beta spots open.

Private beta. We invite teams in batches and never share your email.